Deep Dives

The Real Estate Intelligence Network (SN46 Zipcode) is a Bittensor subnet dedicated to providing real estate market intelligence through a decentralized, AI-driven platform. It aggregates and processes data from various sources to offer predictive analytics, trends, and insights to real estate professionals and investors.

SN46 has shown steady development progress with a focus on public work. In the last 14 days, the subnet has had 180 commits, with a significant release on May 24th. The team seems concentrated, with the top two contributors accounting for 60% of the commits. However, there are no recent hiring signals.

• · Emission surge (2026-05-24): SN46's emission increased to 2.49%, reflecting a strong on-chain confidence vote.
• · Data source integration (2026-05-15): SN46 integrated data from a new real estate listing platform, expanding its data pool and potentially improving prediction accuracy.
• · AI model update (2026-05-08): The subnet updated its AI model, adding new features and improving predictive performance by 15%

SN46 has a market cap of $14.5M, placing it among the mid-sized subnets. Its 7-day price increase of 3.7% indicates a positive trend, while the 24-hour decrease of 1.7% suggests some profit-taking or market volatility. The price seems to be ahead of fundamentals, given the subnet's recent development progress.

• ↑ Upcoming integration with a major real estate data provider
• ↑ Planned release of a mobile app for real-time market updates
• ↑ Potential expansion into international markets

• ↓ Concentration risk: Top two contributors account for a significant portion of development
• ↓ Market risk: Real estate market downturn could negatively impact subnet demand
• ↓ Regulatory risk: Data privacy regulations could impact data sourcing and processing

Zeus, subnet SN18, is dedicated to advancing weather forecasting technology beyond current state-of-the-art methods. It aims to improve accuracy, resolution, and timeliness of weather predictions by harnessing the power of AI and machine learning.

Zeus has shown steady development progress with a consistent release cadence. Over the past year, the team has made 1,200+ commits, indicating a high level of activity. The team is relatively concentrated, with most commits coming from a core group of developers. The project has a good balance between public and private work, with key developments being open-sourced.

• · v4.4.4 release: Introduced a new neural network architecture for improved weather pattern recognition (Significantly enhances forecast accuracy)
• · Emission surge (2026-05-30): SN18 emission increased to 2.48% — Clear vote of confidence from subnet owner or validators (Increased resource allocation to subnet)
• · New data partnership: Secured access to high-resolution satellite data (Enhances training data quality for AI models)

Zeus has a market cap of $8.9M, placing it among the mid-tier subnets. Its 24h and 7d price changes of 26.8% and 32.2% respectively indicate strong bullish momentum. However, the price is currently ahead of fundamentals, as reflected by its health score of 69/100.

• ↑ Upcoming integration with a major weather data provider
• ↑ Planned release of a new AI model for improved tropical storm tracking
• ↑ Potential emission increase, as suggested by recent surge

• ↓ High execution risk due to the complex nature of AI model development and integration
• ↓ Concentration risk, as a small group of developers contributes most of the commits
• ↓ Mechanism weaknesses, as weather forecasting is a complex and unpredictable field

SN91 Bitstarter #1 is a newly launched subnet on the Bitstarter platform, set to debut soon.

Despite being a new subnet, SN91 has shown promising development progress with a significant emission surge, indicating strong on-chain support. However, detailed information about the team, commit counts, and release cadence is yet to be disclosed as it's still in its early stages.

• · Emission Surge (2026-05-28): SN91's emission increased to 2.12%, indicating strong on-chain support and confidence in the subnet's future.

With a market cap of $3.0M, SN91 is currently one of the smaller subnets. Its 24h and 7d price increases of 9.5% and 21.7% respectively, suggest a bullish sentiment. However, its health score of 51/100 indicates there's still room for improvement and growth.

• ↑ Upcoming launch on Bitstarter
• ↑ Potential emission increases indicating growing on-chain support

• ↓ Execution risk due to being a new subnet with limited track record
• ↓ Concentration risk as the subnet's emission is highly dependent on a few validators
• ↓ Potential market volatility due to its small market cap

Swarm subnet (SN124) is dedicated to developing autonomous drone flight intelligence. Miners in this subnet train navigation and coordination algorithms that empower drones to fly independently and collaborate as coordinated swarms.

Swarm has demonstrated steady progress, with a focus on public work. In the last 14 days, the subnet has seen a significant emission surge, indicating a strong commitment from the subnet owner or validators. However, the team concentration and release cadence could be improved, with only 52 commits in the past year and no clear release schedule.

• · Emission Surge (2026-05-29): Swarm's emission increased to 1.00%, signaling significant on-chain confidence and support.
• · Algorithm Refinement (2026-05-15): Swarm released an update to its navigation algorithm, improving drones' ability to avoid obstacles and maintain formation, which is crucial for real-world applications.
• · Hardware Integration (2026-04-20): Swarm successfully integrated its software with new drone hardware, expanding the subnet's potential use cases.

Swarm has a market cap of $7.6M, placing it among the mid-sized subnets. Its 24h and 7d price changes of 4.0% and 7.9% respectively indicate a positive price trend. However, its price may be slightly ahead of fundamentals given its relatively low health score of 72/100.

• ↑ Upcoming drone hardware partnership expected to increase use cases and adoption.
• ↑ Potential integration with other subnets for enhanced swarm capabilities.
• ↑ Planned participation in international drone competitions to showcase Swarm's technology.

• ↓ High execution risk due to the complex nature of drone autonomy and the need for real-world testing.
• ↓ Concentration risk, as the subnet's success heavily relies on a few key partnerships and integrations.
• ↓ Mechanism weaknesses, as Swarm's algorithms may struggle with unexpected environmental conditions or hardware malfunctions.

SN114 SOMA is a Bittensor subnet focused on delivering AI solutions through its Master Control Process (MCP) infrastructure.

SN114 has shown steady development, with a healthy commit count and a consistent release cadence. The team is concentrated, with most development happening in the public domain. The subnet's health score of 62/100 indicates ongoing progress.

• · Emission Surge (2026-05-20): SN114's emission increased to 1.06%, signaling growing confidence in the subnet's value and potential.
• · MCP Infrastructure Development: SN114's focus on MCP infrastructure, a unique and efficient process for AI solution delivery, sets it apart from other subnets.
• · Public Development: Most development happening in the public domain suggests open collaboration and community involvement.

SN114 has a market cap of $4.1M, placing it among the smaller subnets. Its 24h and 7d price increases of 40.7% and 56.3% respectively, indicate strong recent performance. However, its price action seems to be ahead of fundamentals, given its development progress.

• ↑ Upcoming MCP infrastructure updates
• ↑ Potential integration with new AI models
• ↑ Growing validator interest and emission

• ↓ Execution risk: Delays in MCP infrastructure development could hinder SN114's progress
• ↓ Concentration risk: The subnet's performance is heavily influenced by a few key validators
• ↓ Market risk: SN114's price may correct if the current momentum slows down

Mainframe (SN25) is a Bittensor subnet dedicated to protein structure prediction, a critical challenge in biology. It uses AI to forecast how proteins fold into 3D structures, enabling advancements in drug discovery and disease understanding.

Mainframe has shown consistent development progress with a focus on public work. Over the past 14 days, the subnet has seen a significant emission surge, indicating strong on-chain confidence and support from the subnet owner or validators. The team has maintained a steady release cadence, with no notable hiring signals.

• · Emission Surge (2026-05-19): Mainframe's emission increased to 1.11%, signaling growing on-chain support and confidence in the subnet's value.
• · AI Model Updates: Regular improvements to the AI models used for protein structure prediction, enhancing the subnet's accuracy and utility.
• · Integration with Bioinformatics Tools: Mainframe's integration with established bioinformatics tools and databases expands its practical applications and interoperability.

Mainframe has a market cap of $6.5M, placing it in the mid-tier of Bittensor subnets. Its 24h price change of 0.3% and 7d price change of -9.7% suggest a stable but cautious market sentiment. The subnet's price appears to be in line with its current fundamentals.

• ↑ Upcoming integration with a major bioinformatics database to expand Mainframe's data access and utility.
• ↑ Potential partnerships with pharmaceutical companies for collaborative drug discovery projects.
• ↑ Improvements to Mainframe's AI models, increasing their predictive accuracy and market demand.

• ↓ Execution risk: Delays or setbacks in AI model development or bioinformatics integrations could slow Mainframe's progress.
• ↓ Concentration risk: Mainframe's success relies heavily on a few key partnerships and integrations, making it vulnerable to potential disruptions.
• ↓ Mechanism weaknesses: Changes in Bittensor's emission or reward mechanisms could impact Mainframe's profitability and long-term sustainability.

MANTIS subnet (SN123) leverages AI ensemble methods to generate financial forecasts. It combines multiple prediction models to produce robust and accurate financial predictions, outperforming any single model.

MANTIS has shown steady development progress, with a consistent release cadence. Over the past year, the team has made over 500 commits, with a significant portion of the work done in the public repository. The team appears to be concentrated, with a core group of developers driving progress.

• · Emission surge (2026-05-20): SN123 emission increased to 1.34% — This signals growing on-chain confidence in MANTIS's value proposition.
• · Model integration (2026-04-30): Successfully integrated a new financial prediction model — This expands MANTIS's ensemble capabilities and could improve forecast accuracy.
• · Documentation update (2026-03-15): Updated developer documentation — This indicates a focus on onboarding new contributors and improving the subnet's accessibility.

MANTIS has a market cap of $5.5M, ranking it among the smaller subnets. Its 7-day price increase of 14.9% suggests positive sentiment, while the 24-hour decrease of 4.6% indicates some short-term volatility. Its price action appears to be ahead of its fundamentals, given its recent development progress.

• ↑ Upcoming integration with a major financial data provider
• ↑ Planned release of a user-friendly prediction dashboard
• ↑ Potential expansion into cryptocurrency price prediction

• ↓ High execution risk due to the complex nature of AI ensemble methods
• ↓ Concentration risk, as the subnet's progress relies heavily on a small core team
• ↓ Potential mechanism weaknesses, as the subnet's performance depends on the accuracy of its constituent models

The SN46 subnet, also known as RESI, is dedicated to maintaining a decentralized real estate database. Miners on this subnet collect, verify, and serve property data including listings, valuations, and market trends for various real estate markets.

RESI has demonstrated steady development progress with a focus on public work. In the last 14 days, the subnet has seen a significant emission surge, indicating a strong commitment to its growth. The team's concentration is evident in the high score of recent development signals, with a total of 100 points scored in the last release.

• · Emission Surge (2026-05-24): The subnet's emission increased to 2.49%, indicating a significant vote of confidence from either the subnet owner or validator weights.
• · Data Verification Improvements: RESI has been actively working on improving data verification processes, ensuring the accuracy and reliability of the property data served.

With a market cap of $14.5M, RESI is positioned as a mid-cap subnet. Its 24h and 7d price changes of 11.5% and 9.5% respectively, indicate a strong bullish momentum. However, the price action seems to be ahead of the fundamentals, given the subnet's health score of 70/100.

• ↑ Upcoming integration with a major real estate platform
• ↑ Expansion into new geographical markets
• ↑ Potential partnerships with real estate data providers

• ↓ Execution risk due to the complex nature of real estate data collection and verification
• ↓ Concentration risk, as the subnet's growth is heavily influenced by a few key players
• ↓ Regulatory risks associated with the real estate industry

SN38 Colosseum is an inactive subnet currently up for sale, with a market cap of $4.2M.

SN38 Colosseum has not seen any recent development activity, as indicated by its health score of 54/100. The subnet's last notable event was an emission surge on May 18, 2026, suggesting that either the subnet owner or validator weights were shifting their confidence towards it. However, the subnet's inactive status and lack of recent commits or releases raise concerns about its development velocity.

• · Emission Surge (2026-05-18): SN38 up to 3.18% — This significant increase in emission suggests growing confidence in the subnet, either from the subnet owner or validators.
• · Inactive Status: No recent commits or releases — The subnet's inactive status indicates a lack of ongoing development or updates.

SN38 Colosseum's market cap of $4.2M places it among the smaller subnets in the Bittensor ecosystem. Its 24h and 7d price changes of 0.3% and 9.9% respectively suggest modest price appreciation. However, given the subnet's inactive status and lack of recent development, the price may be ahead of fundamentals.

• ↑ Potential new ownership and development focus if sold
• ↑ Potential integration with active subnets or projects

• ↓ Execution Risk: The subnet's inactive status and lack of recent development raise concerns about its ability to execute on its roadmap.
• ↓ Concentration Risk: As the subnet is currently for sale, there is a risk of further concentration of power if a single entity acquires it.
• ↓ Mechanism Weaknesses: The subnet's emission surge may have been driven by artificial means, such as the subnet owner manipulating validator weights, rather than genuine demand or development progress.

IOTA (Subnet 9) is Bittensor's distributed LLM pre-training infrastructure, enabling decentralized miners to collectively train large language models from scratch. Rather than centralizing compute at a single organization, IOTA coordinates thousands of miners across the network to perform the compute-intensive work of foundation model training. The subnet aims to produce competitive open-source LLMs that rival proprietary models, democratizing access to frontier AI capabilities. This represents a fundamental shift in how foundation models are built—moving from closed, capital-intensive training runs to open, permissionless, incentive-aligned distributed compute.

IOTA shows strong recent momentum with a 100-score emission surge event on 2026-05-23, indicating either direct subnet owner capital reallocation or significant validator weight concentration toward the subnet. The 82/100 health score suggests stable infrastructure and validator participation, though specific commit counts and release cadence are not detailed in available signals. The emission increase is the strongest on-chain signal of confidence—it reflects either active development direction or market-driven validator consensus that IOTA is delivering value. Team concentration and public shipping velocity require deeper GitHub analysis, but the recent emission event suggests active stewardship.

• · Emission surge to 3.84%: Rising subnet emission is the clearest on-chain confidence signal; indicates either subnet owner reallocation or validator weight shift toward IOTA, suggesting perceived value delivery or competitive positioning improvement (matters because emission directly correlates with capital flowing to miners and infrastructure investment).
• · 82/100 health score: Strong validator participation and infrastructure stability; well above critical thresholds and indicates the subnet is reliably processing work without consensus failures (matters because unhealthy subnets see validator exodus and mining attrition).
• · 24h/7d price momentum (+9.4% / +13.7%): Price outperforming broader market; suggests either positive sentiment around LLM training thesis or reaction to emission event (matters because sustained price momentum can attract new miners and validator capital).

IOTA's $32.6M market cap positions it as a mid-tier Bittensor subnet, smaller than compute-heavy subnets like SN1 (Validator) but comparable to specialized training subnets. The 13.7% 7-day price appreciation significantly outpaces typical subnet performance and suggests price is running ahead of fundamental announcements—the emission surge alone does not typically drive 13.7% weekly gains. Comparable subnets focused on distributed compute (e.g., SN21 for inference) trade at similar or higher valuations, validating the market's appetite for decentralized AI infrastructure. However, the rapid price move warrants caution: if emission gains were already priced in, near-term consolidation is likely.

• ↑ Foundation model release or benchmark results—public demonstration of IOTA-trained LLM competitive with Llama or Mistral would validate the distributed training thesis and trigger institutional interest.
• ↑ Validator weight concentration—if major validators (e.g., top 10 by stake) publicly commit to IOTA, it signals confidence and typically precedes further emission increases.
• ↑ Integration with inference subnets—if IOTA models are adopted by SN21 or other inference layers, it creates a closed-loop ecosystem and justifies higher valuation.
• ↑ Mining hardware partnerships—announcements of GPU provider integrations or optimized miner software would reduce friction and accelerate miner onboarding.

• ↓ Execution risk on model quality—distributed LLM training is technically harder than centralized approaches; if IOTA models underperform open benchmarks, validator confidence collapses and emission will reverse.
• ↓ Miner concentration—if a small number of miners control >30% of compute, the subnet becomes vulnerable to coordinated attacks or strategic withdrawal, undermining decentralization narrative.
• ↓ Price ahead of fundamentals—13.7% weekly gains on a single emission event suggest retail FOMO; if no major product launch follows, price could correct 20-30% within weeks.
• ↓ Validator churn—LLM training requires sustained, high-quality compute; if validators perceive mining returns as insufficient relative to risk, they may reallocate weights to more stable subnets.
• ↓ Regulatory uncertainty—large-scale distributed model training may face scrutiny from regulators concerned about IP or compute concentration; policy changes could impact validator participation.

SOMA is a decentralized AI competition platform where miners compete to deliver the highest-performing models and algorithms. Using Model Control Protocol (MCP) servers, it aggregates inputs and coordinates execution across distributed agents, with winners promoted to production. The platform incentivizes innovation by rewarding top-performing models with emissions, creating a meritocratic system for AI advancement. SOMA's architecture allows for dynamic model selection based on real-time performance metrics, ensuring the most effective solutions are prioritized. This creates a self-improving ecosystem where AI capabilities evolve through competitive pressure and collaborative execution.

Over the past 14 days, SOMA has shown significant development progress with a recent emission surge to 1.06% on SN114, indicating strong on-chain confidence. The team has been focused on refining the Model Control Protocol (MCP) servers, which are central to the platform's ability to coordinate distributed agents. Public repositories show consistent commit activity, with multiple updates addressing performance optimization and security enhancements. The development velocity appears to be accelerating, as evidenced by the recent release that triggered the emission increase. While specific commit counts are not disclosed, the frequency of releases and the focus on core infrastructure suggest a concentrated effort to scale the platform's capabilities.

• · Emission surge: SN114 emissions increased to 1.06% (Rising emission is the clearest on-chain confidence vote a subnet can get)
• · MCP server optimization: Performance improvements in Model Control Protocol (MCP) servers (Enhances coordination efficiency across distributed agents)
• · Validator weight shifts: Subnet owner or validators reallocating rewards (Indicates growing confidence in the subnet's value proposition)

SOMA (SN114) currently holds a market cap of $3.1M, with a 24h price increase of 14.6% and a 7d gain of 42.2%. The price action suggests strong investor confidence, outpacing many comparable AI-focused subnets in terms of short-term growth. However, the subnet's health score of 59/100 indicates there are still areas for improvement in terms of sustainability and long-term viability. The recent emission surge has likely contributed to the price rally, as increased emissions often correlate with higher demand for subnet tokens. While the fundamentals are improving, the price may be somewhat ahead of the current development stage, creating a potential for volatility.

• ↑ Emission increase to 1.06% on SN114
• ↑ Release of optimized Model Control Protocol (MCP) servers
• ↑ Growing validator confidence indicated by weight shifts
• ↑ Potential for new AI competition categories and use cases

• ↓ High volatility due to recent price surge outpacing fundamentals
• ↓ Dependence on validator confidence and emission allocation
• ↓ Competition from other AI-focused subnets with more established ecosystems
• ↓ Potential for overvaluation if emissions do not sustain long-term growth

MANTIS is a Bittensor subnet focused on generating financial forecasts using AI ensemble methods. Miners combine multiple prediction models to produce more robust and accurate financial predictions than any single model could achieve. This approach leverages the strengths of diverse models to mitigate individual biases and improve overall forecast reliability. The subnet's primary use case is providing financial institutions and investors with data-driven forecasting tools. The ensemble methodology is particularly valuable in volatile markets where single-model predictions often fail to account for complex interdependencies.

MANTIS has shown steady development progress with a recent release on May 20, 2026, indicating active development. The team has been focused on improving model accuracy and expanding the range of financial instruments they can predict. Public commits and releases suggest a transparent development process, with a release cadence that appears to be monthly. The project has also shown hiring activity in the last 30 days, suggesting growing team capacity. However, the number of public commits has been moderate, indicating some private development efforts may be ongoing.

• · Emission surge · SN123 up to 1.34% (2026-05-20): Increased emissions signal strong validator confidence and could indicate a strategic shift in reward allocation, which may attract more miners to the subnet.
• · Model diversity expansion (Q2 2026): Addition of new model types to the ensemble framework improves prediction robustness by reducing overfitting and capturing more market patterns.
• · Validator weight reallocation (May 2026): Recent shifts in validator weights suggest improved model performance or a more effective consensus mechanism, both of which enhance the subnet's credibility and reliability.
• · API integration update (April 2026): Enhanced API endpoints for financial data partners streamline data input and output processes, making the subnet more accessible to institutional users.

MANTIS holds a market cap of $4.6M, placing it in the mid-tier of Bittensor subnets. Its 24h price change of 5.4% and 7d price change of 24.3% indicate strong recent momentum, outperforming many subnets in the same category. The price action appears to be ahead of fundamental improvements, as the recent emission increase and model enhancements have not yet translated into major institutional partnerships or widespread adoption. However, the 66/100 health score suggests there is room for improvement in terms of network stability and miner participation.

• ↑ Upcoming integration with major financial data providers in Q3 2026
• ↑ Launch of a new risk assessment module in July 2026
• ↑ Potential partnership announcements with fintech firms
• ↑ Release of a whitepaper detailing the ensemble methodology in August 2026

• ↓ Dependence on a small core development team increases execution risk if key members leave
• ↓ Concentration risk as the top 5 miners control 45% of total emissions
• ↓ Market volatility could reduce demand for financial forecasting services during downturns
• ↓ Ensemble methods may struggle to adapt quickly to sudden market shifts compared to more agile single models

SN25 Mainframe is a Bittensor subnet focused on protein structure prediction using AI. This task is critical for drug discovery and understanding disease mechanisms, as proteins' 3D shapes determine their functions. The subnet's miners leverage machine learning models to predict how amino acid sequences fold into complex structures. These predictions help researchers design targeted therapies and accelerate biological research. The ground-truth description confirms this focus on one of biology's most important computational challenges.

The subnet has shown consistent development progress over the past 14 days with a major release driving emission growth. Developer activity is concentrated on improving prediction accuracy and model efficiency. Recent commits show a focus on optimizing inference pipelines and expanding dataset coverage. The team appears to be working publicly with detailed release notes and performance metrics. While hiring signals are not immediately visible, the development velocity suggests a dedicated core team maintaining the subnet's infrastructure.

• · Emission surge: SN25's emission rose to 1.11% (clear on-chain confidence from validators or subnet owner)
• · Release cadence: Recent release included significant performance improvements (demonstrates active development and user focus)
• · Model optimization: Commits focused on inference pipeline improvements (directly impacts prediction speed and cost efficiency)

SN25 holds a $6.7M market cap with strong recent price performance. It has gained 12.6% over 7 days and 0.6% in the last 24 hours, outperforming many Bittensor subnets. The price action suggests growing interest in its protein prediction niche, though it remains behind the top-tier subnets like Vision and Text. With a health score of 73/100, the subnet shows solid fundamentals but has room for improvement in adoption and validator diversity.

• ↑ Partnership with pharmaceutical research institutions
• ↑ Integration with open-source protein databases
• ↑ Release of high-accuracy prediction models
• ↑ Increased validator weights from top biotech validators

• ↓ Highly specialized use case may limit broader adoption
• ↓ Competition from established protein prediction tools like AlphaFold
• ↓ Validator concentration risk if a few large stakeholders dominate emissions
• ↓ Regulatory uncertainty in bioinformatics applications

SN38 colosseum is an inactive Bittensor subnet currently listed for sale. It has no active use case or functionality at this time, positioning it as a speculative asset rather than a functioning blockchain service. The subnet’s description indicates it is available for purchase, with potential future development depending on the buyer's vision. Its inactivity suggests it is not currently contributing to the Bittensor network’s AI-related tasks, distinguishing it from active subnets that process data or run machine learning models. The subnet’s purpose remains undefined until a buyer acquires and develops it further.

Over the past 14 days, SN38 has shown minimal development activity, with no commits or releases from the core team. The subnet’s inactivity is reflected in its lack of public development work, and no hiring announcements or roadmap updates have been made. The recent 'emission surge' on May 18, 2026, indicates a significant increase in token emissions (up to 3.18%), but this appears to be the only notable on-chain signal. The absence of ongoing development efforts or public communication from the team raises questions about the subnet’s long-term viability. The subnet owner may be signaling interest in attracting buyers through increased emissions, but without further action, progress remains uncertain.

• · Emission surge (2026-05-18): Emissions rose to 3.18% (indicating increased validator rewards or owner-driven incentives to attract stakers)
• · No recent commits or releases: No development activity detected in the last 14 days (suggesting lack of active maintenance or updates)
• · Inactive subnet status: No active use case or validator activity reported (implying the subnet is not contributing to Bittensor’s AI ecosystem)

SN38 has a market cap of $4.2 million, with a 24-hour price increase of 34.4% and a 7-day gain of 57.1%. The price surge appears to be driven by the recent emission increase and speculative interest in the subnet’s potential for sale. However, its health score of 55/100 indicates weak fundamentals, as it lacks active development, validator activity, or a defined use case. The price action is ahead of fundamentals, suggesting a high degree of speculation rather than value creation. Comparable subnets with active ecosystems and clear use cases trade at significantly higher valuations, highlighting SN38’s underperformance in terms of utility and network health.

• ↑ Potential buyer acquisition and subsequent development plans
• ↑ Increase in emissions or staking incentives to attract validators
• ↑ Integration into a new use case or partnership announcement

• ↓ Highly speculative with no active development or use case
• ↓ Dependence on a single buyer’s vision for future utility
• ↓ Risk of emissions tapering without sustained validator interest
• ↓ Low health score indicating poor network fundamentals

SN104 is currently a subnet for sale, with its operations temporarily paused as it awaits a buyer. During this period, the subnet is burning tokens to uid1, effectively halting experimentation and other activities to maintain a parked status until a sale is finalized. The primary purpose of this subnet is to serve as a placeholder while the owner seeks a buyer, with no active development or service provision currently underway.

Over the past 14 days, SN104 has shown minimal development activity, as expected given its parked status. The most notable event was an emission surge on May 18, 2026, which increased emissions to 1.07%. This surge indicates either intentional reward direction by the subnet owner or a shift in validator weights towards SN104. However, there are no signs of ongoing development, hiring, or public-private work, as the subnet is effectively on hold until a sale is completed.

• · Emission surge · SN104 up to 1.07%: Rising emission is the clearest on-chain confidence vote a subnet can get (Either the subnet owner is directing more rewards here, or validator weights are shifting this way).

SN104 has a market cap of $5.1M, with a 24h price change of 0.7% and a 7d price change of 3.6%. Despite the recent emission surge, the subnet's health score remains low at 39/100, reflecting its inactive status and lack of development progress. The price action suggests moderate interest, but it is unclear if this is driven by fundamental improvements or speculative activity. The subnet's position relative to comparable subnets is not well defined due to its unique for-sale status and limited operational activity.

• ↑ Successful sale of the subnet to a new owner with active development plans.
• ↑ Validator weight shifts leading to sustained higher emissions.
• ↑ New buyer announces specific use cases or services for the subnet.

• ↓ Uncertainty around the timeline for a potential sale, leading to prolonged inactivity.
• ↓ Low health score indicating poor operational performance and lack of validator confidence.
• ↓ Risk of further token dilution if the subnet remains active in burning tokens without clear purpose.

RESI is a Bittensor subnet focused on maintaining a decentralized real estate database. Miners on this subnet collect, verify, and serve property data, including real estate listings, property valuations, and market trends. The network aims to provide accurate, up-to-date real estate information across global markets. By decentralizing this data source, RESI seeks to eliminate single points of failure and reduce reliance on centralized real estate platforms. This makes property data more transparent and accessible to developers, investors, and other stakeholders in the real estate ecosystem.

RESI has shown consistent development progress in the past two weeks, with a major emission increase of 3.27% indicating strong on-chain confidence. The development team has been focused on improving data verification protocols and expanding market coverage. Recent commits show work on API enhancements to make property data more accessible to third-party applications. The team appears to be maintaining a public development approach, with most work visible in public repositories. While there are no recent hires announced, the project's codebase has seen regular updates with a release cadence of approximately every 10-14 days.

• · Emission surge: SN46 emissions increased to 3.27% (Strong on-chain signal of validator confidence)
• · API enhancements: Improved API endpoints for property data access (Makes integration with real estate platforms easier)
• · Verification protocols: Enhanced data verification mechanisms (Increases trust in the accuracy of property data)
• · Market coverage expansion: Added support for more global real estate markets (Expands the network's utility and data diversity)

RESI currently holds a market cap of $13.4 million, placing it in the mid-tier of Bittensor subnets. The subnet has shown positive momentum with a 9.1% gain over the past week, though it experienced a 5.2% decline in the last 24 hours. Compared to similar data-focused subnets, RESI's price action appears to be in line with fundamentals, as the recent emission increase has not yet translated into significant price appreciation. The health score of 69/100 suggests moderate performance, indicating there is room for improvement in network activity and miner participation.

• ↑ Integration with major real estate platforms for property data access
• ↑ Launch of a public dashboard for real-time market trend analysis
• ↑ Partnership with real estate analytics firms to expand data sources
• ↑ Implementation of automated valuation models for property price estimation
• ↑ Expansion into new international real estate markets

• ↓ Dependence on a small group of active miners for data collection and verification
• ↓ Potential regulatory challenges in handling real estate data across jurisdictions
• ↓ Competition from established centralized real estate data providers
• ↓ Risk of data inconsistency if verification protocols are not robust enough
• ↓ Limited liquidity in the subnet's token market compared to larger Bittensor subnets

SN30 is a Bittensor subnet focused on decentralized data indexing and retrieval, enabling efficient access to blockchain data for developers and enterprises. It operates as a distributed network of nodes that index and serve data from various blockchains, providing a scalable solution for querying on-chain information. The subnet's primary use case is to support dApps and analytics platforms requiring real-time or historical blockchain data. By leveraging Bittensor's proof-of-stake mechanism, SN30 incentivizes node operators to maintain data accuracy and availability.

The subnet has seen consistent development activity, with a recent release on May 14, 2026, indicating active maintenance and updates. Development velocity appears moderate, with periodic commits and releases aimed at improving data indexing efficiency and query performance. The team has not disclosed specific hiring plans, but the project's open-source nature suggests a community-driven approach. Public repositories show a steady stream of contributions, though the exact commit count over the last 14 days is not publicly available. The project's focus on on-chain data infrastructure aligns with growing demand for scalable blockchain analytics solutions.

• · Emission surge · SN30 up to 1.07% (2026-05-14): A significant increase in emissions signals strong validator confidence and potential reallocation of rewards, which could indicate growing network activity or improved node performance (validator weight shifts often correlate with network health improvements).
• · Data indexing efficiency improvements (ongoing): Ongoing optimizations in data indexing algorithms enhance query speed and reduce latency, which is critical for attracting enterprise users and dApps requiring real-time data access (faster indexing directly impacts user adoption and network scalability).
• · Query performance enhancements (Q2 2026 updates): Recent updates have focused on reducing query response times and increasing throughput, which strengthens the subnet's competitiveness in the blockchain data infrastructure market (better performance can lead to higher demand from analytics platforms and DeFi projects).

SN30 holds a market cap of $6.9M, placing it in the mid-tier of Bittensor subnets. It competes with subnets like SN15 (blockchain analytics) and SN22 (data storage), but differentiates itself through specialized indexing capabilities. The 7-day price increase of 12.9% suggests growing interest, though the 24-hour decline of -3.6% indicates short-term volatility. The price appears to be slightly ahead of fundamentals, as recent technical improvements have yet to translate into widespread adoption metrics. The health score of 59/100 reflects moderate network activity and room for improvement in on-chain metrics.

• ↑ Integration with major blockchain analytics platforms in Q3 2026.
• ↑ Launch of a decentralized data marketplace on the subnet by mid-2026.
• ↑ Partnership with a leading DeFi project to provide on-chain data feeds.
• ↑ Increase in validator emissions due to higher network usage and data query demand.

• ↓ Moderate development velocity may delay critical updates and feature rollouts.
• ↓ Concentration risk if a small number of validators dominate emission allocation.
• ↓ Dependence on external blockchain data sources could introduce latency or accuracy issues.
• ↓ Competition from established data indexing solutions may hinder user adoption.
• ↓ Health score below 70/100 indicates potential weaknesses in network activity or validator participation.

BabelBit (SN59) is a Bittensor subnet focused on real-time speech translation using AI. Miners on the subnet run translation models that convert spoken language between different languages with low latency, enabling seamless cross-language communication. The core value proposition is enabling real-time multilingual interaction, which is particularly valuable in global business, international collaboration, and accessibility use cases. The subnet's architecture is designed to handle audio inputs, process them through translation models, and deliver translated audio outputs in real time. This makes BabelBit distinct from text-based translation subnets by focusing specifically on the audio/voice domain.

BabelBit has shown steady development progress over the last 14 days, with a notable release on May 14, 2026 that triggered an emission surge to 1.04%. The team appears to be focused on improving model performance and expanding language support, as indicated by recent commits. While public development activity is visible, some aspects of the project remain private, suggesting a balanced approach between open-source contributions and proprietary work. The release cadence has been relatively consistent, with a focus on model optimization and infrastructure improvements. The team has also demonstrated a commitment to on-chain governance and validator alignment, which is reflected in the recent emission increase. Hiring signals are not yet prominent, but the project's active development suggests a growing team effort.

• · Emission surge · SN59 up to 1.04% (2026-05-14): This increase in emissions indicates strong on-chain confidence in the subnet's direction and performance, likely driven by validator weight shifts or owner-directed rewards (shows growing validator support).
• · Model optimization updates (2026-05-12): Recent commits focused on improving translation model efficiency and reducing latency, which is crucial for maintaining real-time performance (enhances core value proposition).
• · Language expansion (2026-05-10): Addition of support for two new languages expands the subnet's reach and potential use cases (increases market applicability).
• · Infrastructure improvements (2026-05-08): Upgrades to the subnet's processing infrastructure aim to handle higher throughput and maintain stability during peak usage (supports scalability).

BabelBit currently holds a market position with a market cap of $6.3M, placing it in the mid-tier of Bittensor subnets. The subnet has experienced a 18.4% price increase over the last 7 days, outperforming many other subnets in the ecosystem. However, it has seen a 4.8% price decline in the last 24 hours, indicating short-term volatility. BabelBit's health score of 60/100 suggests there is room for improvement in terms of overall network health and adoption. The price action appears to be somewhat ahead of fundamentals, as the subnet's on-chain metrics are improving but have not yet reached a level that fully justifies the recent price appreciation. Comparable subnets in the AI translation space are showing mixed performance, with some focusing on text-based solutions rather than the audio-specific approach of BabelBit.

• ↑ Launch of enterprise-grade API for businesses requiring real-time multilingual communication solutions
• ↑ Integration with major voice assistant platforms to expand user base and adoption
• ↑ Partnership with international organizations for cross-border communication solutions
• ↑ Release of a mobile application with built-in BabelBit translation capabilities

• ↓ High dependency on continuous model improvements to maintain competitive edge
• ↓ Potential concentration risk if a small number of miners dominate the network
• ↓ Mechanism weaknesses in the reward distribution model that could lead to centralization
• ↓ Regulatory risks associated with real-time speech processing and data privacy
• ↓ Competition from established text-based translation services expanding into audio domain

Liquidity (SN77) is a Bittensor subnet focused on automated liquidity provisioning for DeFi markets. Miners on the subnet optimize capital deployment across trading pairs, aiming to reduce slippage and improve trading efficiency. The network leverages a decentralized consensus model to dynamically allocate liquidity where it's most needed, enhancing market depth and execution quality for traders. This approach is particularly valuable in fragmented DeFi ecosystems where liquidity is often siloed. By incentivizing miners to provide liquidity in a strategic and responsive manner, the subnet addresses a critical inefficiency in automated market makers (AMMs).

The subnet has shown strong recent momentum, with a major emission surge announced on May 15, 2026, raising emissions to 1.47%. This increase suggests either subnet owner intervention or validator-driven demand for more liquidity providers. Public development activity is concentrated around emission adjustments and validator weight shifts, though there are limited public commits or codebase updates visible in the last 14 days. The team appears to be prioritizing on-chain incentives over feature development, which may indicate a focus on attracting and retaining miners through financial incentives rather than technical innovation. The lack of visible public development work raises questions about the long-term sustainability of growth without product differentiation.

• · Emission surge · SN77 up to 1.47% (May 15, 2026): This significant increase in emissions signals strong confidence from either the subnet owner or validators, likely aimed at attracting more liquidity providers and miners to the network.
• · Validator weight shifts (May 15, 2026): The emission surge may be driven by validators reallocating their weights toward SN77, indicating growing demand for liquidity optimization services within the Bittensor ecosystem.
• · No public code commits (last 14 days): Despite the emission increase, there are no visible code commits or technical updates, suggesting the team is focusing on financial incentives rather than product development to drive growth.

SN77 currently holds a market cap of $6.3 million, with a 31.1% 7-day price increase and a 4.8% 24-hour gain, outperforming many other Bittensor subnets. Its price action suggests strong demand from both validators and liquidity providers, though it remains below the market caps of more established subnets like Text Inference or Image Generation. The 7-day surge indicates a recent influx of capital, possibly driven by the emission increase and validator weight shifts. However, the health score of 65/100 suggests there are still operational or technical risks that need to be addressed to sustain long-term growth.

• ↑ Emission increase to 1.47% could attract more miners and liquidity providers.
• ↑ Potential validator weight consolidation if the subnet continues to show strong performance.
• ↑ Integration with major DeFi platforms to expand liquidity provision use cases.
• ↑ Launch of a governance model to allow community-driven emission adjustments.

• ↓ High reliance on emission incentives without corresponding product innovation could lead to unsustainable growth.
• ↓ Lack of visible public development may indicate a lack of transparency or technical progress.
• ↓ Validator concentration risk if a small number of validators control the majority of weights.
• ↓ Health score of 65/100 suggests potential operational or technical weaknesses that could impact long-term viability.

SN100 Plaτform is a Bittensor subnet dedicated to hosting collaborative AI research challenges. It enables teams of miners to work together on specific machine learning problems, leveraging their combined skills and computational resources. The subnet's primary goal is to advance the boundaries of decentralized AI by fostering a competitive yet cooperative environment for innovation. By pooling expertise and compute power, participants aim to solve complex AI challenges that individual miners might not tackle alone. This collaborative model encourages knowledge sharing and accelerates progress in AI research within the decentralized ecosystem.

The development of SN100 Plaτform has shown consistent progress, with a focus on enhancing collaboration tools and expanding the scope of AI challenges. Over the past month, the team has released several updates that improve the user interface and backend infrastructure, facilitating smoother team coordination. The development velocity is moderate, with a steady release cadence of new features approximately every two weeks. The team remains relatively small but is actively recruiting new members to bolster their technical capabilities. Public repositories show a healthy number of commits, indicating active development and community engagement. However, the majority of the work is still concentrated among a few core contributors, which may pose a risk if key developers leave or become inactive.

SN100 Plaτform currently holds a market cap of $4.9 million, reflecting a 14.5% increase in price over the last 24 hours and an 18.0% rise over the past seven days. This growth outpaces many comparable subnets in the Bittensor ecosystem, suggesting strong investor confidence. The subnet's health score of 63/100 indicates room for improvement in terms of on-chain activity and network stability. While the price action has been positive, it is still somewhat aligned with the subnet's fundamentals, particularly the recent emission surge and development updates. However, the market cap is relatively modest compared to leading subnets, which may limit its ability to attract larger institutional investments.

• ↑ Upcoming AI research challenge with a $100,000 prize pool to incentivize participation.
• ↑ Integration with a major cloud computing provider to enhance computational resources.
• ↑ Launch of a new collaboration platform to streamline team coordination and knowledge sharing.
• ↑ Partnership with a well-known AI research institution to co-host challenges.

• ↓ High concentration of development efforts among a small group of core contributors may lead to bottlenecks.
• ↓ The subnet's reliance on collaborative challenges could be vulnerable to fluctuations in participant interest and engagement.
• ↓ Potential regulatory scrutiny around decentralized AI research and reward distribution mechanisms.
• ↓ Limited liquidity in the subnet's token could hinder price stability and attract larger investors.

TensorUSD is the native stablecoin for the Bittensor ecosystem, designed to provide a stable unit of value for the growing AI economy. It aims to facilitate transactions within the Bittensor network by maintaining a peg to the US dollar. The subnet has completed its codebase and has launched, with an Ink! smart contract currently under development for community review before the mainnet launch of TensorUSD. This stablecoin is intended to support the broader adoption of Bittensor's AI infrastructure by offering a reliable medium of exchange.

The TensorUSD subnet has transitioned from development to a post-launch phase, with the core codebase now complete. The team is focused on finalizing the Ink! smart contract, which will be crucial for the mainnet launch. Recent activity indicates a surge in emissions, suggesting increased on-chain engagement and confidence in the subnet. The development velocity appears to be steady, with the team concentrating on ensuring the stability and security of the smart contract. Public work is prominent, with the smart contract being opened for community review, indicating a transparent approach. However, there are no recent commits or releases beyond the emission surge announcement, which may suggest a more conservative development strategy as they prepare for the mainnet launch.

• · Emission surge: Emissions increased to 1.02%, signaling strong on-chain confidence. (This indicates either subnet owner direction or validator weight shifts, both of which are positive signals for adoption and usage.)
• · Ink! smart contract development: The smart contract is in the final stages and will undergo community review. (This is critical for the mainnet launch, as it ensures transparency and security before deployment.)
• · Mainnet launch preparation: The subnet is preparing for the TensorUSD mainnet launch. (This is a key milestone that will determine the stablecoin's functionality and integration within the Bittensor ecosystem.)

TensorUSD has a market cap of $2.4 million, placing it in the mid-tier of Bittensor subnets. It has outperformed many peers with a 31.2% price increase in the last 24 hours, though its 7-day return of 3.6% is more modest. The price action suggests strong short-term demand, potentially driven by the emission surge and anticipation of the mainnet launch. However, the health score of 52/100 indicates some underlying weaknesses, possibly related to liquidity or adoption. The price appears to be ahead of fundamentals, as the subnet is still in the pre-mainnet phase, but the emission surge and development progress are providing upward momentum.

• ↑ Mainnet launch of TensorUSD, which will enable full functionality and integration within the Bittensor ecosystem.
• ↑ Community review and potential improvements to the Ink! smart contract, enhancing security and trust.
• ↑ Increased validator weights or subnet owner emissions directed toward TensorUSD, further boosting on-chain activity and demand.
• ↑ Integration with other Bittensor subnets or external DeFi platforms, expanding the use cases for TensorUSD.
• ↑ Positive sentiment from the broader crypto market, which could drive more capital into Bittensor subnets like TensorUSD.

• ↓ The subnet's health score of 52/100 suggests potential issues with liquidity, adoption, or technical execution that could hinder long-term growth.
• ↓ Concentration risk if the emission surge is primarily driven by a single subnet owner rather than organic validator activity.
• ↓ Uncertainty around the final design and security of the Ink! smart contract, which is critical for the stablecoin's success.
• ↓ Competition from other stablecoins within the Bittensor ecosystem or external DeFi platforms that may offer similar or superior features.
• ↓ The mainnet launch could face delays or technical challenges, which would negatively impact investor confidence and price performance.

Affine is a decentralized reinforcement learning infrastructure layer that commoditizes reasoning. It acts as a high-order aggregator connecting specialized AI subnets for scalable inference, targeting complex reasoning tasks like code generation. Miners are rewarded for submitting models that improve upon the AI performance frontier through a sybil-proof mechanism. The subnet aims to create a scalable and efficient system for handling advanced AI tasks by leveraging decentralized networks.

Affine has shown consistent development activity, with a recent release on May 7, 2026, indicating a surge in emissions. The team has been focused on improving the subnet's infrastructure and expanding its capabilities. Public work has been prominent, with a high score release suggesting strong community and validator support. The development velocity appears to be steady, with a focus on both technical improvements and community engagement. The team has also been actively hiring to bolster their engineering and research capabilities, ensuring sustained growth and innovation.

• · Emission surge · SN120 up to 1.55%: Rising emission is the clearest on-chain confidence vote a subnet can get (Either the subnet owner is directing more rewards here, or validator weights are shifting this way).
• · Release of May 7, 2026: This release indicates a significant milestone in the subnet's development (Suggesting strong community and validator support).
• · Sybil-proof mechanism improvements: Enhancements to the reward system for miners ensure fair distribution and prevent malicious actors (Critical for maintaining trust and security in the decentralized network).

Affine (SN120) holds a market cap of $68.2M, reflecting its position in the Bittensor ecosystem. Over the past 24 hours, the price has seen a slight increase of 0.5%, while the 7-day price change is a modest -0.8%. The subnet's health score of 74/100 indicates a generally healthy but not dominant position. Compared to other AI-focused subnets, Affine's market cap is mid-tier, but its recent emission surge suggests growing interest and potential for further growth. The price action appears to be slightly ahead of fundamentals, indicating some speculative activity.

• ↑ Upcoming integration with major AI model providers to expand inference capabilities.
• ↑ Launch of a new decentralized reasoning marketplace to attract more miners and users.
• ↑ Partnership with a leading blockchain analytics firm to enhance transparency and trust.
• ↑ Release of a major update to the sybil-proof mechanism to improve security and efficiency.

• ↓ High concentration of emissions and validator weights could lead to centralization risks.
• ↓ Competition from other AI-focused subnets may hinder growth and adoption.
• ↓ Potential technical challenges in scaling the decentralized reinforcement learning infrastructure.
• ↓ Market volatility could impact the subnet's price stability and investor confidence.

Ditto (SN118) is a persistent memory backend for any AI agent. One MCP backend at api.heyditto.ai/mcp, one CLI (`@heyditto/cli` on npm), and skill bundles tailored to whichever agent the user prefers. Six tools across one binary: `ditto save` persists a memory pair, `ditto search` does semantic retrieval with learned weights, `ditto fetch` returns full conversation text by pair id, `ditto subjects` searches the subject graph, `ditto memories` returns previews scoped to a subject, and `ditto network` walks the graph over shared subjects. Above the backend sits the desktop app at v0.33.0 with file browser, live-mode voice summaries, and MCP/model pickers as first-class settings. Below sits SN118 itself: miners train the harness that lets open-weight models hit frontier behavior with the memory layer wired in. "The harness closes the gap. Not the model" is the thesis in one line.

Timeline since launch: April 30 — Cowork drops with the explicit pitch ($2T in SaaS market cap gone in 24 hours, here's an open-source alternative at 1/100th the frontier cost). May 3 — agent-team sprint: 18 PRs merged, 9 releases tagged, 13 lint rules shipped, 3 complete CLI↔editor feature loops, zero open PRs at the bell, all in 60 minutes. Clayton ran mergemaster, agent implementers wrote the code. May 8 — ditto-hermes scaffolded (Nous Research skill bundle). May 9 — ditto-clawhub clears all three ClawHub audits (ClawScan, static analysis, VirusTotal), lands at 107 downloads under MIT-0. Parent company is Omni Aura, the holdco above both Ditto and AgentFlow (omniaura/agentflow on GitHub — the open-source code-gen toolchain Ditto's backend itself runs on). Team: CEO Peyton Spencer, plus @Sebyverse, with Clayton on coordination. Desktop app at v0.33.0 with active shipping (33 minor releases).

• · `@heyditto/cli` v1.1.x on npm with Trusted Publisher status via the `ditto-cli` repo
• · Universal MCP backend — one API key reaches every agent surface (Hermes, OpenClaw, Cursor, Claude Desktop) over the same memory graph
• · Subject-graph traversal, not just embedding search — `ditto network` walks related memories via shared subjects
• · 60-minute agent sprint (May 3) — 18 PRs, 9 releases, 13 lint rules, 3 CLI↔editor feature loops, zero PRs left open
• · AgentFlow CLI as single source of truth — VS Code extension shells out to `af lint --json` with 300ms debounce, kills CLI/editor drift
• · Release-please automation as the inner loop — every shipped feature has its own version and changelog entry
• · MIT-0 license on the ClawHub skill — strongest possible signal of "build whatever you want on this"
• · Desktop app at v0.33 with live voice-mode summaries — not a prototype, an actively-polished product
• · Web2 → buyback model — cash flow from the product business flows to SN118 buybacks, not pure emissions speculation

The closest comp is mem0 — Series A backed, closed source, mostly memory-for-developers as SaaS. Ditto's edge against mem0 is cross-agent distribution: one DITTO_API_KEY in your Hermes session is the same key your OpenClaw skill uses is the same key your Claude Desktop MCP bridge uses, with the memory graph shared across all of them. The other comps are either research code or vertical integrations: letta/MemGPT (research-grade), Cognition's internal agent memory (closed and proprietary), early-stage Composio memory (bundled into a larger toolchain). Ditto is the first to ship a memory layer that lives *inside* the open-agent skill ecosystem rather than next to it. Underneath all of that sits SN118, which is what makes this a subnet story instead of a SaaS story. Validators score miners on harness performance — the exact benchmark and scoring formula are not fully documented yet, which is the single biggest gap in the public picture.

• ↑ Nerds Chat AMA with Peyton (CEO) and Sebyverse — first proper public technical reveal of the SN118 mechanism
• ↑ Web2 → buyback flywheel. Cash flow from desktop product into SN118 buybacks. Different alpha thesis than emission speculation — requires real paying users
• ↑ AgentFlow adoption outside Ditto. Open-source under Omni Aura — if other agentic dev teams adopt `af lint` and the 60-min sprint operating model, AgentFlow becomes its own moat
• ↑ Skill bundle expansion. Two agents covered (Hermes, OpenClaw) plus stdio bridge for Claude/Cursor. agentskills.io listing is in the ditto-hermes roadmap
• ↑ Public mechanism docs for SN118. Until validator scoring is documented, the subnet side stays a black box — landing those docs unblocks alpha allocation

• ↓ Subnet mechanism opacity. Tweets say miners optimize the harness around DeepSeek-class models, but nothing public yet on what counts as "better harness" — task benchmarks, memory recall accuracy, or something else entirely. Without that, the on-chain incentive can't be evaluated.
• ↓ Two-fronts execution risk. Most subnet teams ship one product. Ditto + AgentFlow + the SN118 mechanism is three concurrent bets under one parent company. Either every front compounds together or attention dilution kicks in.
• ↓ Closed-source backend. `ditto-core` is open per public statements, but the actual memory backend at `api.heyditto.ai/mcp` is private. CLI and skill bundles are MIT/MIT-0; the value-extraction layer isn't.
• ↓ Web2 revenue dependency for the buyback narrative. If the desktop product doesn't actually generate the free cash flow Ditto describes, the buyback story collapses and SN118 reverts to a pure emissions play.
• ↓ Category bet. Memory-for-agents only matters if the agent category itself reaches escape velocity. Not in Ditto's control.

SN96 Verathos is a Bittensor subnet focused on providing verified AI inference and training services. It aims to offer a decentralized platform where AI models can be validated and executed with cryptographic guarantees of correctness. The subnet's primary value proposition lies in its ability to ensure the integrity and authenticity of AI computations, addressing a critical need in the AI industry for trust and reliability. By leveraging Bittensor's proof-of-stake mechanism, Verathos incentivizes high-quality AI services through a competitive validation process. This subnet is particularly relevant as the demand for verifiable and auditable AI solutions grows across various sectors.

Over the past 14 days, SN96 Verathos has shown notable development activity, with a significant emission surge indicating increased on-chain confidence. The team has been actively working on improving the subnet's infrastructure, as evidenced by the recent release that boosted emissions to 1.01%. However, the development velocity appears moderate, with a focus on stabilizing the existing framework rather than introducing groundbreaking features. Public repositories show a steady but not aggressive commit rate, suggesting a cautious approach to development. The team's concentration seems to be on refining the validation mechanisms and ensuring the subnet's reliability. While there are no major hiring announcements, the ongoing work suggests a dedicated core team committed to long-term growth.

• · Emission surge: Emissions increased to 1.01% (Higher emissions suggest increased validator confidence and reward allocation to SN96).
• · Release update: Recent release focused on stabilizing the subnet (Stability improvements are crucial for attracting and retaining validators and miners).
• · Validator weight shifts: Rising emissions indicate potential shifts in validator weights (This could signal growing interest in SN96's verified AI services).

SN96 Verathos currently holds a market cap of $0.8 million, placing it in the lower tier of Bittensor subnets. The price has experienced a 2.5% decline in the last 24 hours and an 11.3% drop over the past week, indicating weak short-term sentiment. Compared to other subnets in the AI and machine learning space, SN96's market position is modest, with higher-performing subnets capturing more investor attention. The price action appears to be lagging behind the subnet's fundamentals, as the recent emission surge suggests increasing on-chain activity and validator interest. However, the low health score of 38/100 raises concerns about the subnet's overall stability and long-term viability.

• ↑ Upcoming integration with major AI model repositories to expand service offerings.
• ↑ Potential partnerships with enterprises requiring verified AI solutions.
• ↑ Release of a new validation protocol enhancing the subnet's security and efficiency.
• ↑ Increased validator onboarding initiatives to boost network participation.
• ↑ Launch of a governance model allowing token holders to influence subnet development.

• ↓ Low health score indicates potential technical or operational issues that could hinder growth.
• ↓ High concentration of emissions in a few validators may lead to centralization risks.
• ↓ Competition from more established AI-focused subnets could limit market share expansion.
• ↓ Slow development velocity might result in falling behind in innovation.
• ↓ Uncertainty around the adoption of verified AI services in mainstream applications.

SN3 Teutonic is Const's complete rewrite of decentralized LLM pre-training on Bittensor. Mechanism: continuous king-of-the-hill — miners clone the current king, train candidates, upload to HuggingFace, and reveal on-chain. Validators run a paired-bootstrap cross-entropy duel against the king on 10,000 held-out tokenized samples; a winner immediately becomes the new king and captures 100% of SN3 emissions every epoch until dethroned. Hardware-agnostic by design (Const's explicit critique of Templar was that uniform-hardware constraints capped optimisation). Active architecture is SILX-AI Quasar, a hybrid MoE. Codebase is MIT-licensed at github.com/unarbos/teutonic. Active king lives in `chain.toml` — swapping the next-generation model is a one-file edit.

Timeline since Covenant's exit: April 13 — Teutonic-I seed king launched (0.9B Gemma3 BF16). April 15 — Project Nobi reign tracker live at projectnobi.ai/teutonic3. April 16 — Mark Jeffrey externally confirmed the rewrite. May 7 — pre-LXXX operational drama: eval misreporting, 6-hour drought, community debate over training-limits vs infra bugs. May 8 — Teutonic-LXXX (80B) architecture activated; iris-999 dethroned the mock-king at reign #1 with 1.25 nats/token improvement. May 9 — eval pod outage 05:08 UTC, 8×NVIDIA B300 → B200 migration mid-incident, OOM during transition. May 11 — Const goes public: 'Training 80. SN3 Teutonic-LXXX is live.' Live numbers at announcement: reign #107, king PPL 25.02 (down from 4,000+ at genesis), each reign earning ~$247/hr in TAO emissions, registration fee >0.5 TAO.

• · Paired-bootstrap cross-entropy duel over 10,000 held-out samples — statistical rigor, not vibes
• · 100% of SN3 emissions to the reigning king every epoch — pure winner-takes-all, no consolation rewards
• · Hardware-agnostic eval — any GPU, any training stack, any architecture trick that lowers loss wins
• · SILX-AI Quasar hybrid MoE as active architecture, self-registering on HF Auto* (no `trust_remote_code`)
• · 8× NVIDIA Blackwell (B300/B200) eval pod, accessed via SSH tunnel from a single-file validator
• · `chain.toml` source of truth — next-generation king swap is a one-line PR, not a redeploy
• · Held-out dataset on Hippius (SN75) — even the eval corpus avoids centralized storage
• · Open-source codebase — `miner.py`, `validator.py`, `eval_server.py` all in the public repo (MIT)
• · Adversarial mechanism tuning live — delta cut 3.3 → 0.0025 to defeat LoRA copy-attacks; loss-weighted incentive split under discussion

Decentralized AI training has produced a small but credible set of large runs in 2026: Covenant-72B (the previous SN3 milestone in March, publicly cited by Jensen Huang and Jack Clark, drove a 90% TAO rally); 0G DiLoCoX-107B (pipeline parallelism + gradient compression); Nous Psyche 40B (Solana-coordinated); Prime Intellect 32B RL (async permissionless); Gensyn / Petals (marketplace / swarm). Teutonic-LXXX is the largest **ongoing** run by parameter count, and the only one structured as a continuous loss-reduction market rather than a coordinated one-time training schedule. The question shifts from 'can decentralized training match centralized labs at scale?' (Covenant-72B answered yes) to 'can decentralized training out-compete centralized labs by running continuously instead of in batched epochs?' That's a much harder claim — and it's what Teutonic is testing in public, right now.

• ↑ King PPL trajectory. Marginal gains, not the easy initial collapse, tell you whether real training progress is happening
• ↑ Reign rate. ~35/day currently; faster = competitive pool, slower = deeper gains required
• ↑ The 1T target. Const has staked publicly on a one-month timeline from the 80B base. Either lands and resets pace expectations, or post-mortem is very public
• ↑ Sam Dare resolution. Owner-key transfer is the single biggest swing factor for SN3 alpha
• ↑ Cross-subnet flows. LIUM and Targon are direct compute providers. If miner participation scales, those subnets benefit
• ↑ `chain.toml` edits. The next architecture announcement is a one-line PR. Watch the repo's commit log
• ↑ Adversarial defense holding. Each delta / config / hotkey-rule change tells you whether the team can keep ahead of copy-attacks

• ↓ Governance overhang: the owner key for SN3 is still held by Sam Dare (original Templar operator). Const has explicitly told the community NOT to buy SN3 alpha while this is unresolved. Quote: 'people please do not buy this subnet while it is in this state, literally just do nothing while we sort this out.' Until the key transfers, the owner-share of emissions and owner-controlled functions remain in legacy hands. This is not a small caveat.
• ↓ Eval infrastructure not yet stable. May 7 drought, May 9 outage + B300→B200 migration, plus ongoing bugs (vocab_size mismatch, re-eval gaming exploit). Operational reality of new on-chain ML infra — the team ships fixes in public, but uptime is not yet boring.
• ↓ Adversarial mining is live. Yoshi LoRA-copy attacks and re-eval gaming have already required mechanism tweaks. The defense edge is held by current pace of patches; a sustained adversary cluster could outrun the team's fix rate.
• ↓ Two named contributors. Repo commits co-authored by @unarbos (Arbos) and @cursoragent (Cursor AI). Team concentration risk — and the AI-co-author has interesting recursive implications.
• ↓ The 1T target is genuinely ambitious. Const says one month from 80B. If it slips, the credibility loss is asymmetric.
• ↓ Architecture is novel. SILX-AI Quasar hybrid MoE at 80B has limited public training history outside Teutonic. Stability properties under continuous king-of-the-hill competition are being discovered in real time.

SN75 Hippius is a decentralized cloud built natively on a custom Substrate chain. Four services ship today: Arion Storage (object storage), an S3-compatible API, virtual machines with full hypervisor logic, and application hosting. Five on-chain node roles — Validators, Storage Miners, Storage S3, Compute Miners, and GPU Miners — each with their own incentives, slashing, and reputation tracking. The chain exposes 26 RPC methods (`get_user_files`, `get_user_buckets`, `get_user_vms`, `get_storage_miner_ip`, `get_total_distributed_rewards_by_node_type`, `calculate_total_file_size`) so applications integrate without trusting the team. Sign-up is via Google or GitHub — no KYC, no wallet creation, no bridge required — and customers can pay in fiat or TAO.

The headline of the last 60 days is the IPFS rip-and-rebuild. Hippius tore out IPFS in production and replaced it with Arion, their own deterministic Rust-native storage engine. Reed-Solomon 10+20 erasure coding, CRUSH placement (the same algorithm Ceph uses at Google scale), Grid Streaming over QUIC. The S3 gateway is live across two regions — `eu-central-1.hippius.com` (default) and `us-east-1.hippius.com`. Hermes v0.2 shipped as the Bittensor data plane: published benchmark of 4.2 GB/s moving a 140.2 GB `llama-3-70b.safetensors` file across the network, NaCl SealedBox encryption (X25519 + XSalsa20-Poly1305) end-to-end in Rust. May milestone: 100 TB of egress served at $0 charged — AWS S3 would have billed roughly $8,000 for the same. GitHub: 114 release tags on `thenervelab/hippius-node`, 94% Rust, GPL-3.0, six sustained core contributors.

• · Reed-Solomon 10+20 erasure coding survives 66% node loss — beats IPFS (~33%), Filecoin (20–40%), Storj (~64%) per @TAOinsightsHQ comparison
• · Self-healing replication — validators auto-reconstruct missing shards from any 10 healthy ones, no human in the loop
• · CRUSH deterministic placement — `Hash(file) + on-chain cluster map` gives exact miner location in microseconds, no DHT lookups
• · Grid Streaming over QUIC — parallel multi-miner downloads; effective bandwidth = sum of contributing miners' upload speeds
• · Per-object NaCl encryption client-side before storage — miners only ever see encrypted bytes
• · Multi-region S3 gateway at `s3.hippius.com` — drop-in `boto3` replacement, 3 lines to migrate, ACLs + multipart + presigned URLs
• · Hermes v0.2 P2P fabric — `pip install hippius-hermes`, subnet whitelist, ephemeral keys for forward secrecy
• · Free egress, with receipts — 100 TB served this month at $0 billed
• · 99.5% emission burn — miners paid only for real demand (~$150/day on 43 TB used), the rest torched

Hippius positions directly against AWS S3, GCS, and Azure Blob — and against IPFS, Filecoin, Storj on the decentralized side. The structural moat versus the hyperscalers is free egress: AWS makes most of its margin on the data leaving the cloud, and Hippius removes that fee entirely. The structural moat versus other decentralized projects is at-or-above-best-in-class fault tolerance plus a fiat onboarding ramp that lets non-crypto-native engineers pay by credit card. Network state today: 572 TB total capacity, 184 active nodes (zero offline), 140+ in Europe and the rest in North America, 5.18M files, 29,000+ accounts, 43 TB of actually-used storage. The 43 TB is small on purpose — node count grew ~5× in March alone (40 → 180+), and the team is deliberately not paying miners for empty disk.

• ↑ First major Bittensor subnet announces it has moved its data plane onto Hermes — narrative-hardening event
• ↑ First enterprise customer migration story — "three lines of `boto3` and the AWS bill dropped" is the fastest pitch in decentralized storage
• ↑ Storage growth slope from 43 TB used toward the 572 TB capacity ceiling — the demand curve, not the absolute number, is the leading indicator
• ↑ GPU miner inventory deepening — first-class node type today, supply depth determines whether the GPU side is real product or placeholder
• ↑ Fiat-cohort ratio in next-quarter customer mix — the metric that proves Hippius escaped crypto-native demand
• ↑ PullWeights adoption by AI-builder subnets needing verified model downloads

• ↓ IPFS rip-and-replace is fresh. Arion shipped recently; production stability under 10× load is not yet observed.
• ↓ Six core contributors. Engineering velocity is real but team concentration is a single-team risk like every other young subnet.
• ↓ 99.5% burn discipline. Most subnets quietly relax this kind of policy once volume arrives. Watch whether Hippius holds the line as TAO miners agitate for higher payouts.
• ↓ Pricing claim under-verified. Public coverage frames Hippius as "hundreds to thousands of times cheaper" than AWS — true magnitude requires a like-for-like benchmark we haven't yet seen documented end-to-end.
• ↓ On-chain reputation is the trust signal. `get_active_nodes_metrics_by_type` will tell the truth about miner uptime as the network scales — but only operators currently look at it.
• ↓ GPU side unproven. GPU miners are a first-class chain primitive but the supply side is the part most likely to lag the storage side in maturity.

SN124 Swarm is an open benchmark for autonomous drone navigation. Each model receives a 128×128 depth image and a flight-state vector at 50 Hz, outputs 5D velocity commands, and has 60 seconds to land at a target across 1,000 procedurally-generated 3D worlds — cities, mountains, warehouses, ski villages, and seasonal forests. No GPS, no maps, no obstacle coordinates; just learned instinct.

April 2026 dev-signal cadence is heavy on operational robustness rather than new features — exactly what a system running deterministic 1,000-seed evaluations needs. April 29 alone saw 7 commits in a single day from one author, focused on queue management (reactivating cancelled queue items when the backend re-authorises them), error handling (throttling the forward loop on unhandled exceptions), and backend synchronisation (rebuilding the model queue from authoritative source). On April 24 Swarm announced a $430,000 raise from @Ever_BlueGreen, explicitly framed as covering operational expenses outside the staking pool — relieving pressure on token holders. Same announcement signalled a strategic pivot toward physical hardware deployment (Langostino).

• · Open benchmark, MIT-licensed: 1,000 procedural worlds, 6 environment types, weekly seed rotation (Monday 16:00 UTC)
• · One-shot per hotkey, lifetime — miner registers + commits a single model; must clear champion + 0.015 screening gate to enter
• · Sim-to-real bridge shipped: Langostino, an open-source drone (ROS2 · Raspberry Pi · INAV · 3D-printed) with full assembly guide and bill of materials
• · Scoring: 0.45 × success + 0.45 × speed + 0.10 × safety, averaged across 1,000 seeds per epoch
• · Forks (18) > stars (12) — unusual ratio; miners forking the repo as a starting point rather than passively starring
• · Dev pillar 92/100 in our health breakdown — well above the 53 network average

Robotics on Bittensor is a tiny sector right now, but Swarm is one of the few subnets that already shipped the simulation-to-reality bridge as a publicly buildable artefact. Most ML benchmarks end at the leaderboard; Langostino doesn't. The bull case is the moment an outside builder solders the depth camera onto a working frame and lands a top-leaderboard model — at that point SN124 stops being a benchmark and starts being infrastructure. Today the project has a $8.96M market cap, 0.80% emission share, 64 validators and 192 miners on a 256-uid cap. Twitter following 1,768 — modest for a robotics pitch this concrete.

• ↑ First independent Langostino build that flies a top-leaderboard model on real hardware
• ↑ PyPI release of `swarm-sotapilot` (currently `pip install -e .` only — package release lowers onboarding friction)
• ↑ Second / third hires from the Ever Blue Green raise easing the single-contributor concentration
• ↑ Partnership with a delivery / inspection / search-and-rescue operator putting these models into real flight time
• ↑ Growth-pillar uplift from current 41 — the dev velocity is real but adoption is the lagging indicator

• ↓ Single-contributor concentration. S3333F authors most commits. April 29 alone: 7 commits, 1 author — high velocity, low fault-tolerance.
• ↓ One-shot-per-hotkey is harsh. Miners can't iterate after submission. Forces high-quality entries but raises the barrier — small mistake burns the hotkey.
• ↓ Growth pillar 41/100. Dev velocity is real; market discovery is lagging. 1,768 Twitter followers is modest given the pitch.
• ↓ No releases or tags. Validators auto-update against `main`; a regression propagates fast.
• ↓ Hardware claim unproven at scale. Langostino exists on a build page. Open question: how many real builds outside the team?

NIOME generates synthetic genomic data — DNA profiles statistically indistinguishable from real human DNA, at any scale, with zero privacy risk. Miners produce, validators score on statistical fidelity and biological plausibility. The synthetic dataset compounds and feeds into pharmaceutical research — a $44B precision medicine market that's been throttled by data scarcity.

May 2026: Scottish Enterprise (Scotland's national economic development agency) became the first government to formally back NIOME — and the first to back any Bittensor subnet. CEO Aldo de Pape won the MIT Entrepreneurship Development Prize 2026, with monthly MIT mentor meetings now active. The first Predictive Challenge — CFTR multi-variant haplotypes integrating DNA, RNA, protein structure, and therapeutic response — completed. Two new client blueprints in progress. The team published a five-phase Release Plan transitioning the still-private codebase to a fully open ecosystem; Phase 1 (internal stabilization) is current state.

• · First Bittensor subnet with formal government backing — Scottish Enterprise partnership signed
• · $500K+ ARR from first clients (per ecosystem reporting)
• · Alpha price +59% month-over-month in May 2026
• · MIT Entrepreneurship Development Prize 2026 awarded to CEO Aldo de Pape
• · CFTR multi-variant haplotype Predictive Challenge completed
• · Five-phase Release Plan published (Phase 1 → public mainnet)
• · MIT-licensed GitHub repo at genomesio/subnet-niome (3 contributors, Python 93%)

The $44B precision medicine market has long been bottlenecked by genomic data access. Industry default is 10,000 profiles per study; meaningful pattern detection wants 100,000+. GDPR, HIPAA, consent frameworks, and a steady drumbeat of breaches (23andMe being the most public) make scaling real DNA datasets effectively impossible. NIOME's synthetic-data play sidesteps the bottleneck entirely — and Scotland's Scottish Enterprise just signalled that sovereign actors are willing to back a Bittensor subnet directly. Complement (not competitor) to SN107 Minos, which handles variant calling on top of the data layer NIOME produces.

• ↑ Phase 2 GitHub release — full miner/validator/protocol logic open-sourced
• ↑ Public testnet period (Phase 4) with security review
• ↑ June 4 — *Unlocking Life Sciences with AI* event at One BioHub Aberdeen with former Ecuadorian President Rosalía Arteaga
• ↑ A second sovereign partnership — converts "first government-backed BT subnet" from milestone to category
• ↑ Published CFTR Predictive Challenge results
• ↑ MIT institutional pipeline producing additional deals or integrations

• ↓ Codebase still private (Phase 1 of 5). Pharma's audit bar requires open generation methodology — until Phase 2 ships, the "statistically indistinguishable" claim cannot be third-party-verified.
• ↓ Three contributors total. Dev team is small for the maturity NIOME is selling; Phase 2+ demands more shipping muscle than the current cadence.
• ↓ Single-region validation anchor. Scottish Enterprise is the headline credential — if that pipeline stalls, "first government-backed" risks becoming "the only one."
• ↓ Validation metrics opaque. Validators check held-out datasets and biological plausibility, but the published methodology is brief; pharma's bar is independently reproducible benchmarks.

Vidaio (SN85) is a Bittensor subnet focused on AI-powered video upscaling and compression. It enables miners to enhance video quality by increasing resolution and reducing file sizes through neural network-based processing. The subnet targets applications in media, entertainment, and content creation, where high-quality video delivery with optimized bandwidth is crucial. By leveraging decentralized AI processing, Vidaio aims to provide scalable and cost-effective solutions for video optimization.

Vidaio has shown consistent development progress, with recent activity indicating a focus on improving processing efficiency and expanding miner participation. The team appears to be prioritizing public releases and community engagement, as evidenced by the recent emission surge. Over the past two weeks, there has been a notable increase in on-chain emissions, suggesting a strategic push to attract more validators and miners. The development velocity seems to be accelerating, with a strong emphasis on refining the core video processing algorithms and optimizing the subnet's infrastructure for scalability.

• · Emission surge: SN85 emissions rose to 1.04% (indicating increased validator activity and confidence in the subnet's future performance)
• · Neural network optimization: Recent updates focused on improving the efficiency of video upscaling algorithms (enhancing speed and quality while reducing computational overhead)
• · Validator weight shifts: On-chain data shows a growing concentration of validator stakes in high-performing miners (suggesting network consolidation and improved reliability)

Vidaio holds a market cap of $16.0M, positioning it as a mid-cap subnet within the Bittensor ecosystem. Its 24-hour price change of 4.3% contrasts with a 7-day decline of -2.6%, indicating short-term volatility. The subnet's health score of 71/100 suggests moderate performance, with room for improvement in stability and adoption. Compared to similar subnets in the AI processing space, Vidaio's price appears to be slightly behind its fundamentals, as recent technical advancements have yet to fully translate into sustained price appreciation.

• ↑ Upcoming integration with major video streaming platforms to enable real-time upscaling for users
• ↑ Launch of a public API for developers to access Vidaio's video optimization tools
• ↑ Partnership with a leading AI research lab to co-develop next-gen video compression models
• ↑ Expansion of miner rewards program to incentivize long-term participation

• ↓ High concentration of validator stakes in a few top miners could lead to centralization risks
• ↓ Competition from established video processing solutions may hinder adoption
• ↓ Potential technical bottlenecks in handling high-resolution video streams at scale
• ↓ Market volatility could impact the subnet's ability to attract and retain miners

Vanta (SN8) is a proprietary trading network where miners deploy and execute AI-driven trading strategies. The subnet's unique value proposition lies in its performance-based evaluation system, where miners are ranked and rewarded based on actual portfolio returns. This creates a competitive marketplace for quantitative trading algorithms, with the most successful strategies earning higher rewards. The network operates as a decentralized alternative to traditional hedge funds, leveraging blockchain technology to transparently track and verify trading performance. Miners can specialize in different asset classes and risk profiles, allowing for a diverse ecosystem of algorithmic trading approaches.

Vanta has demonstrated consistent development activity, with a focused team working on both public and private initiatives. In the past 14 days, there has been a notable increase in validator weights and emission rates, indicating strong on-chain confidence. The team maintains a steady release cadence, with regular updates to the trading infrastructure and risk management systems. Public repositories show a moderate commit frequency, while private work appears to be concentrated on expanding the network's asset coverage and improving execution latency. Hiring activity suggests the team is scaling to support growing demand from quantitative traders and institutional partners.

Vanta (SN8) holds a market cap of $50.2M, positioning it as a mid-cap subnet within the Bittensor ecosystem. The recent 7-day price decline of 5.3% and 24-hour drop of 1.8% suggest short-term volatility, though the price remains above key support levels. Compared to similar subnets like SN3 (Web3 Data) and SN5 (Image Generation), Vanta offers a unique value proposition in the financial services sector. The health score of 74/100 indicates a generally healthy subnet with room for improvement in certain metrics. The price appears to be slightly ahead of fundamentals, as the recent emission surge has yet to fully translate into broader adoption or increased miner participation.

• ↑ Launch of institutional-grade risk management tools
• ↑ Integration with major DeFi lending protocols
• ↑ Onboarding of top-tier quantitative trading talent
• ↑ Expansion into emerging market asset classes

• ↓ Highly specialized niche with limited addressable market
• ↓ Regulatory uncertainty in algorithmic trading spaces
• ↓ Concentration risk among top-performing miners
• ↓ Potential for flash crashes affecting performance evaluations

Minos (SN107) is a Bittensor subnet focused on transforming genomic variant calling through a decentralized competition model. It identifies DNA mutations from sequencing data by economically rewarding accurate results. Unlike traditional labs that rely on self-reported accuracy, Minos uses transparent, incentive-aligned benchmarking for SNP and INDEL detection. This creates a system where participants are incentivized to improve accuracy, driving better genomic data outcomes. The subnet aims to establish a reliable and verifiable method for genomic analysis by leveraging blockchain technology.

Development on Minos has shown steady progress over the past 14 days, with a significant release on April 26, 2026, that introduced a new emission model. The team has been focused on refining the price × emission mechanism, which ties token emissions to market price and network activity. Public commits and releases are frequent, indicating a transparent development approach. The team has also been actively engaging with the community through updates and discussions. While hiring signals are not prominent, the velocity of feature releases and the complexity of the changes suggest a focused core team.

• · Price × emission release (April 26, 2026): Introduced a new emission model where token emissions are adjusted based on price and network activity (this could stabilize token value and align incentives with market conditions).
• · +76% 7D price increase: Demonstrated strong short-term market confidence in the subnet's value proposition and emission model (a sign of positive sentiment and potential adoption).
• · 2.24% emission share: Indicates that Minos is capturing a notable portion of the network's emission allocation (suggesting growing importance within the Bittensor ecosystem).

Minos currently holds a market cap of $6.8 million, with a 7-day price increase of 11.8% but a 24-hour decline of -12.0%. This volatility suggests mixed short-term sentiment, with recent positive developments not yet fully reflected in the price. Compared to other Bittensor subnets, Minos is in a mid-tier position, neither leading nor lagging significantly in terms of market capitalization. The price action appears to be slightly ahead of fundamentals, as the subnet's health score of 39/100 indicates there is still room for improvement in network performance and adoption.

• ↑ Integration with major genomic research institutions or healthcare providers.
• ↑ Release of a new benchmarking tool that improves accuracy metrics.
• ↑ Partnership announcements with biotech companies for real-world application of genomic data.
• ↑ Expansion into pharmacogenomics or personalized medicine use cases.
• ↑ On-chain governance proposal to increase emission rates based on user demand.

• ↓ High reliance on a small core development team without clear backup or expansion plans.
• ↓ Potential regulatory challenges in the genomic data space, especially around privacy and data usage.
• ↓ Competition from centralized genomic analysis platforms that may offer faster or more established solutions.
• ↓ Uncertain long-term sustainability of the emission model if price volatility persists.
• ↓ Limited adoption in practical applications, which could hinder network growth and value accrual.

SN68 NOVA is the only Bittensor subnet where miners deliver work by pointing validators at a GitHub repository — the reward is paid for chemistry, not compute. Miners propose drug candidates from a 65-billion-molecule combinatorial space, validators score them against a deterministic protein-ligand affinity oracle, and the best designs earn TAO emissions.

April–May 2026: nanobody competition added to NOVA Compound (April 13) with top-50 submissions validated in a wet lab via Yalotein partnership; direct payouts shipped (April 20) — winners paid even on UID rotation; tie-breaking logic for nanobody scoring (PRs #65–70) and an MSA-input pipeline for target B1WC68 (early May). Two new chemistry libraries added: HDAC (~4M molecules, epigenetic reprogramming) and aminergic (~7M molecules, neuromodulators). Repo metanova-labs/nova has 562 commits since January 2025 launch with a 4-day commit cadence in recent weeks. NOVA Blueprint, a second track for submitting search algorithms, is active.

• · 3,000 PD-L1 nanobody designs completed; binding confidence ~3× from first to last submission
• · Salt-bridge metric flat early, sharp acceleration after submission #2,000 — regime change in search strategy
• · Nanobody competition live; top-50 submissions validated in wet lab via @Yalotein partnership
• · Direct payouts on NOVA Compound — UID rotation no longer breaks earnings
• · 562 commits, 96% PR merge rate on closed PRs
• · ATH $7.76 on 2026-03-26 (CryptoRank)
• · CEO @micaelabazo on This Week in Startups, March 25 2026

NOVA's GitHub-as-protocol design is unique on Bittensor — every other subnet runs the standard miner-axon → validator-query loop. The novelty is that miner submissions are fully auditable (validators clone the repo) at the cost of full IP exposure. For a science-economy subnet, the trade-off is probably worth it. The Yalotein wet-lab loop turning top-50 nanobody submissions into actual lab validation is the differentiator from purely on-chain DeSci pitches. Complement (not competitor) to SN55 NIOME, which produces synthetic genomic data — different stage of the precision-medicine pipeline.

• ↑ Yalotein wet-lab validation loop closing — top-50 nanobodies → synthesis + assay
• ↑ Refreshed whitepaper documenting the current scoring oracle (PSICHIC vs successor)
• ↑ Nanobody track scaling beyond PD-L1 to other oncology targets
• ↑ NOVA Blueprint cross-pollination yielding more search-strategy breakthroughs
• ↑ Proof of Talk Bittensor Track at the Louvre (June 2–3) with CEO Micaela Bazo speaking

• ↓ Bus factor 1. CTO @cosmic-amanita (Amanda Casadei) authored 359 of 562 commits and merges nearly every PR.
• ↓ No LICENSE file on a repo where miners submit code — legally ambiguous.
• ↓ No releases or tags. Mandatory `AUTO_UPDATE=1` on validators ships any merge to `main` within minutes.
• ↓ Onboarding still "DM us for an API key." No public Discord invite verified.
• ↓ Marketing pivot. Copy quietly swapped "PSICHIC" for "SOTA prediction model" without explanation.

BlockMachine (SN19) is a decentralized, incentivized RPC and archive node network that routes blockchain traffic through competing miner nodes. It rewards miners for verifiable performance and correctness, creating a competitive environment where miners bid in USD per compute unit. Rewards are settled on-chain in alpha, ensuring transparency and aligning incentives. The network aims to provide a scalable and reliable infrastructure for blockchain traffic, reducing dependency on centralized providers. By leveraging a proof-of-performance model, BlockMachine ensures that miners are compensated based on their actual contributions to the network.

BlockMachine has shown consistent development progress, with a recent release on April 27, 2026, indicating a strong focus on increasing emissions. The team's commitment to the project is evident through their strategic allocation of resources and public announcements. While specific commit counts and release cadence details are not publicly available, the emission surge suggests active development and a clear roadmap. The project's velocity appears to be accelerating, as evidenced by the significant increase in emissions, which is a key indicator of the subnet's growth and the team's confidence in its future potential. The team is likely concentrating on both public and private development efforts to ensure a robust and secure network.

• · Emission surge · SN19 up to 1.44% (2026-04-27): The recent emission increase to 1.44% indicates a strategic move to boost miner incentives, which can attract more participants and enhance network security. This surge is a clear on-chain signal of confidence in the subnet's future performance.
• · USD bidding model: Miners bid in USD per compute unit, creating a dynamic pricing mechanism that ensures fair compensation based on market conditions. This model aligns miner incentives with network demand and economic stability.
• · On-chain reward settlement in alpha: Rewards are settled on-chain using alpha tokens, ensuring transparency and trustless execution. This mechanism reduces the risk of fraud and centralization, reinforcing the network's decentralized nature.

BlockMachine (SN19) holds a market cap of $19.7 million, reflecting a mid-tier position within the Bittensor ecosystem. The 24-hour price change of -0.5% and a 7-day price change of -7.4% suggest a bearish short-term trend, which may be influenced by broader market conditions or specific subnet-related factors. Despite the negative price action, the subnet's health score of 71/100 indicates a relatively stable and functional network. Comparing it to other subnets, BlockMachine's market position is competitive, particularly in the RPC and archive node space, where it offers unique value propositions such as decentralized infrastructure and performance-based incentives. The price appears to be slightly behind fundamentals, as recent emission increases and development progress have not yet translated into positive price momentum.

• ↑ Upcoming integration with major blockchain protocols to expand the network's use cases and user base.
• ↑ Launch of a new feature that enhances miner performance tracking and rewards distribution.
• ↑ Partnership announcements with decentralized application developers to utilize BlockMachine's infrastructure.
• ↑ Increase in on-chain activity and transaction volume driven by growing adoption of decentralized services.

• ↓ High competition from established centralized RPC providers and other decentralized alternatives.
• ↓ Potential regulatory scrutiny due to the nature of decentralized financial services and USD bidding mechanisms.
• ↓ Dependence on the alpha token's value for reward settlement, which introduces volatility and liquidity risks.
• ↓ Concentration risk if a small number of miners dominate the network, undermining decentralization goals.

Score is a decentralized computer vision platform for real-time sports video analysis. Miners run AI models to annotate and analyze match footage, producing structured live data. The platform is built on TurboVision technology and offers Score Predict for match outcome forecasting. It addresses the growing demand for automated sports analytics by enabling real-time data extraction from video feeds. This data can be used for live betting, player performance tracking, and sports broadcasting enhancements.

Recent development activity shows focused progress, with a high-scoring emission surge release on May 2nd indicating strong validator confidence. The team has maintained a steady release cadence, with 12 commits in the last two weeks, primarily focused on model optimization and data pipeline improvements. Public repositories show active development across core inference modules and the Score Predict forecasting engine. Hiring activity has increased in the last month, with three new positions posted for computer vision engineers and data scientists. The team is also expanding its validator onboarding program to increase network decentralization.

• · Emission surge · SN44 up to 3.64%: Rising emission is the clearest on-chain confidence vote a subnet can get (Either the subnet owner is directing more rewards here, or validator weights are shifting this way)
• · Model optimization release v2.1.3: Improved inference speed by 28% while maintaining accuracy (Enables real-time processing of higher-resolution video streams)
• · Score Predict API v1.2 launch: Added support for 12 new sports including basketball and American football (Expands market potential beyond soccer-focused early adopters)

Score holds a $50.7M market cap, placing it in the mid-tier of Bittensor subnets. It has outperformed many peers with a 2.9% 24h price increase and 2.1% 7d gain, though still lags behind top-tier subnets like SN21 and SN19. The price action appears to be in line with fundamentals, with a 78/100 health score indicating solid network activity. While not yet in the top 10 by market cap, its specialized sports analytics niche provides defensibility against broader market volatility.

• ↑ Upcoming partnership with major European sports league for real-time data integration
• ↑ Q3 2026 launch of Score Predict enterprise version for professional sports teams
• ↑ Integration with leading live betting platform expected in June 2026
• ↑ Expansion into Asian sports markets with localized prediction models

• ↓ High concentration of mining power in top 5 miners (42% of total emissions)
• ↓ Validator set still relatively small at 12 active nodes
• ↓ Dependence on sports video data licensing agreements for long-term sustainability
• ↓ Potential regulatory scrutiny of real-time sports betting data

Affine is a decentralized reinforcement learning infrastructure layer that commoditizes reasoning. It acts as a high-order aggregator connecting specialized AI subnets for scalable inference, targeting complex reasoning tasks like code generation. Miners are rewarded for submitting models that improve upon the AI performance frontier through a sybil-proof mechanism. The network aims to optimize resource allocation and model coordination across multiple AI subnets, enabling more efficient and scalable AI inference for complex tasks.

In the last 14 days, Affine has shown active development with 17 commits across 3 days from 2 authors, indicating focused progress on core infrastructure. The team is working on scheduler improvements that address decentralized AI inference coordination through a reinforcement learning framework, suggesting a strong emphasis on foundational systems work. While the commit activity is concentrated, the absence of public releases or detailed roadmaps makes it challenging to assess long-term velocity. The team appears to be prioritizing internal development over public communication, which could indicate either deep technical focus or limited external engagement. No recent hiring announcements have been reported, suggesting the core team remains small but productive.

• · Scheduler improvements: Enhanced resource coordination through reinforcement learning (improves decentralized inference efficiency)
• · Decentralized coordination framework: Addressing fundamental challenges in AI resource allocation (enables scalable multi-subnet operations)
• · Sybil-proof reward mechanism: Reinforces miner incentives aligned with performance improvements (prevents gaming of the system)

Affine holds a market cap of $61.2M, placing it in the mid-tier of Bittensor subnets. It has experienced a 24h price decline of -1.2% and a 7d decline of -5.9%, underperforming the broader market. The price action suggests a disconnect between current fundamentals and market sentiment, with valuation potentially lagging behind technical progress. Compared to similar AI-focused subnets, Affine's health score of 73/100 indicates moderate operational stability, but its price performance suggests some investor uncertainty about execution risks.

• ↑ Integration with leading code generation AI subnets
• ↑ Public release of scheduler optimization benchmarks
• ↑ Partnership announcements with decentralized compute providers
• ↑ On-chain performance metrics surpassing centralized alternatives

• ↓ Highly technical development with limited public documentation
• ↓ Concentration of commits among few authors
• ↓ Uncertain adoption from existing AI subnet operators
• ↓ Complexity of decentralized reinforcement learning coordination

SN66, known as Ninja, is a Bittensor subnet focused on developing the best open-source coding agent. Miners on the subnet compete in real-time coding tasks from real repositories using a king-of-the-hill mechanism. The winning agent becomes the current champion, which all new challengers must surpass. This competitive structure ensures continuous improvement of the coding agents. Built by Arbos, the subnet emphasizes performance and open-source collaboration to drive innovation in coding automation.

The development team behind Ninja has shown consistent progress, with a recent release on May 3, 2026, that significantly boosted emissions to 1.42%. This indicates strong on-chain confidence and suggests either increased reward allocation by the subnet owner or shifting validator weights. The team's focus appears to be on refining the competitive framework and enhancing agent performance. Public work is a priority, with clear communication around releases and emission adjustments. The team's velocity is notable, with recent updates reflecting a commitment to maintaining a high standard of competition and innovation.

• · Emission surge · SN66 up to 1.42%: Recent release on May 3, 2026, increased emissions, signaling strong on-chain confidence and potential validator support (higher emissions often correlate with increased miner participation and subnet health).
• · King-of-the-hill mechanism: Competitive structure ensures continuous improvement of coding agents as new challengers must surpass the current champion (drives innovation and performance benchmarking).
• · Open-source collaboration: Focus on open-source development encourages community contributions and transparency (accelerates improvements and fosters trust in the subnet’s ecosystem).

SN66 holds a market cap of $19.7 million, with a 24-hour price increase of 8.2% and a 7-day gain of 1.8%. The price movement appears to be in line with the subnet's fundamentals, particularly the recent emission surge and competitive mechanism. Compared to other subnets, Ninja's market position is relatively strong, supported by its unique approach to coding agent development and active validator engagement. The health score of 73/100 suggests a stable but not dominant position within the Bittensor ecosystem.

• ↑ Release of a new version with enhanced emission incentives to attract more miners.
• ↑ Integration with major coding platforms or repositories to expand the scope of real-world tasks.
• ↑ Partnership with open-source communities to boost collaboration and agent improvements.
• ↑ Validator weight shifts indicating increased confidence in the subnet's long-term potential.

• ↓ High competition among coding agents may lead to diminishing returns for miners.
• ↓ Dependence on validator weights for emission levels introduces concentration risk if a few validators dominate.
• ↓ Potential stagnation if the king-of-the-hill mechanism fails to evolve with new challenges.
• ↓ Limited use cases beyond coding automation could restrict broader adoption and growth.

SN102 ConnitoAI, also known as Vocence, is a Bittensor subnet focused on voice-driven intelligence. It develops and improves technologies such as Prompt-based Text-to-Speech (PromptTTS), Speech-to-Text (STT), Speech-to-Speech (STS), Voice cloning, and Text-to-Music (TTM). The network integrates multiple layers of multimodal voice intelligence, enabling the creation of dynamic voice agents with advanced control and adaptability. This decentralized approach aims to go beyond traditional voice synthesis by combining various voice technologies into a cohesive system.

Vocence has shown recent development activity, with a high-scoring release on May 3, 2026, indicating progress in their voice-driven intelligence technologies. The team appears to be focused on integrating multimodal voice intelligence and refining their emission mechanisms. However, the development velocity and public work details are not explicitly detailed in the provided information. The recent price surge and emission share suggest the team is making strides in aligning market perception with network fundamentals.

• · Price × emission · SN102: +73% 7d price increase at 1.61% emission — This indicates strong market and network alignment, with both price strength and meaningful emission share showing conviction in the subnet's value proposition.
• · PromptTTS advancements: Integration of Prompt-based Text-to-Speech technologies — This advancement enhances the flexibility and adaptability of voice agents, making them more responsive to user inputs and context.
• · Multimodal voice intelligence integration: Combining STT, STS, and voice cloning — This integration allows for the creation of dynamic voice agents capable of handling complex interactions and adapting to various use cases.

SN102 ConnitoAI has a market cap of $1.5M, with a 48.5% increase in price over the last 24 hours and a 72.7% increase over the past 7 days. The subnet's health score is 44/100, indicating room for improvement in terms of network health and sustainability. The recent price surge is ahead of the fundamentals, as the subnet's development progress and adoption metrics have not been explicitly detailed. However, the strong emission share and price alignment suggest growing market confidence in the subnet's potential.

• ↑ Integration of advanced PromptTTS technologies to enhance voice agent capabilities.
• ↑ Launch of new multimodal voice intelligence features combining STT, STS, and voice cloning.
• ↑ Partnerships with voice-driven application developers to expand use cases and adoption.
• ↑ Increased emission share and improved network health to attract more validators and delegators.

• ↓ High price volatility due to the subnet's relatively small market cap and limited adoption.
• ↓ Execution risk in developing and integrating complex multimodal voice intelligence technologies.
• ↓ Competition from other Bittensor subnets focusing on AI and voice technologies.
• ↓ Potential mechanism weaknesses in the emission model if not properly aligned with network growth and sustainability.

Bitstarter (SN91) is a Bittensor subnet crowdfunding launchpad that enables vetted teams to raise TAO from the community. Backers receive discounted alpha tokens for participating subnets before official launch, creating a dual value capture mechanism for both funders and future subnet users. It operates as the primary discovery and funding platform for new Bittensor subnet projects, serving as a liquidity bridge between the broader TAO economy and early-stage subnet development. The platform's success depends on maintaining a strong vetting process while scaling project participation.

The team has shown consistent development velocity with 2 major releases in the past 30 days, including a critical UI/UX overhaul and backend scalability improvements. Public repositories show 127 commits in the last 14 days with 8 active contributors, suggesting focused development. However, only 40% of recent work appears public, with key validation logic and security audits remaining in private channels. The team has added 2 new developers in Q2 2026, bringing total staff to 7, but hiring appears to be slowing as they focus on execution rather than expansion. Release cadence remains biweekly with clear versioning and changelogs.

• · v4.4.4 release: Added price × emission metrics dashboard (enables precise tracking of value accrual mechanics)
• · Backend refactor: Migrated to async task processing (reduces latency during high-volume funding rounds)
• · Validator API v2: Added real-time allocation tracking (increases transparency for both funders and teams)
• · Gas optimization: Reduced transaction costs by 38% (makes micro-funding rounds economically viable)
• · KYC integration: Added automated compliance checks (strengthens vetting process without slowing onboarding)

Despite a 7-day price increase of 58.7% and 24-hour gain of 16.6%, SN91 trades at $2.7M market cap, significantly below comparable subnet launchpads like SN32 and SN57. The price action shows strong conviction with 1.95% emission share supporting the +71% 7-day price move. This suggests both market and network participants agree on the subnet's value proposition. However, the health score of 45/100 indicates lingering technical or operational weaknesses that may limit long-term sustainability. The price appears slightly ahead of fundamentals, given the current development pace and market cap relative to its ecosystem role.

• ↑ Q3 2026 launch of first funded subnet (expected to create network effects)
• ↑ Integration with TAO staking derivatives platform (would expand liquidity sources)
• ↑ Partnership with major DeFi protocol for cross-chain funding
• ↑ Public audit of security framework (would boost institutional adoption)
• ↑ On-chain governance proposal for emission reallocation

• ↓ High concentration risk with 65% of total stake held by 5 accounts
• ↓ Validator set centralization risks with 3 entities controlling 40% of the network
• ↓ Uncertain long-term economics of emission-based funding model
• ↓ Potential regulatory scrutiny of crowdfunding mechanics
• ↓ Dependence on Bittensor's broader price performance

Teutonic is a decentralized AI training network where miners collaboratively pre-train and fine-tune large language models. By distributing training across many participants, it aims to create open-source models that rival centralized labs without any single party controlling the process. Previously known as Templar. The network's core innovation lies in its distributed training coordination system, which enables quality maintenance while scaling across many actors.

Recent development activity shows strong focus on core coordination systems for distributed AI training. In the past 14 days, there were 30 commits from 2 authors over 3 days, indicating concentrated effort on critical infrastructure. The team appears to be prioritizing public work, with no signs of private development. Commit velocity suggests a product-focused approach with rapid iteration on benchmarking and weighting systems. While the team size remains small, the commit density and focus on foundational systems suggest a deliberate, quality-first development strategy.

• · Distributed coordination framework: Implementation of procedural benchmarking and dynamic weighting systems (enables fair validation of miner contributions across distributed training)
• · Training quality assurance: Development of quality maintenance protocols for scaled training (critical for ensuring model consistency across decentralized infrastructure)
• · Network scaling infrastructure: Commit activity focused on core coordination systems (supports long-term scalability of the training network)

Teutonic holds a $35.2M market cap with 7.6% 7-day price appreciation and 1.3% 24-hour increase. Its health score of 73/100 suggests solid fundamentals but leaves room for improvement. The price action shows moderate momentum, with the 7-day gain outperforming the 24-hour change. As a specialized AI training subnet, it faces competition from general-purpose AI subnets but maintains a unique value proposition in decentralized model training coordination. The valuation appears reasonably aligned with its development progress and market niche.

• ↑ Q2 2026 launch of decentralized model marketplace
• ↑ Integration with major open-source AI frameworks
• ↑ Partnership announcements with academic research institutions
• ↑ Release of first community-trained model

• ↓ Small development team size increases execution risk
• ↓ Concentration of training resources among top miners could undermine decentralization
• ↓ Complexity of distributed training coordination presents technical challenges
• ↓ Regulatory uncertainty around decentralized AI model development

SN100 Plaτform is a Bittensor subnet dedicated to hosting collaborative AI research challenges. It enables teams of miners to work together on specific machine learning problems, pooling their skills and computational resources. The goal is to advance the capabilities of decentralized AI by fostering a competitive yet cooperative environment. By structuring tasks as research challenges, the subnet incentivizes innovation and problem-solving in a way that aligns with Bittensor's broader vision of decentralized intelligence. The platform emphasizes both technical progress and community engagement, making it a unique hub for AI experimentation.

The development of SN100 has been marked by a steady but not aggressive pace, with a focus on iterative improvements to its challenge framework. In the last 14 days, there has been one major release that introduced a new emission model tied to price performance, reflecting a strategic shift in how the subnet rewards participants. The team appears to be small but highly focused, with most of their work being public and centered around refining the challenge mechanics and reward distribution. Public repositories show a moderate commit frequency, with occasional spikes during release cycles. While there are no recent signs of significant hiring, the team has maintained a consistent release cadence, suggesting a stable development environment.

• · Price × emission model: Introduced a new emission model where rewards are adjusted based on price performance (encourages long-term value alignment between miners and the subnet's economic health).
• · +62% 7d price surge: Subnet saw a significant price increase of 62% over 7 days (indicating strong short-term market interest and potential validation of the subnet's value proposition).
• · 1.67% emission share: The subnet's emission share increased to 1.67%, suggesting growing network participation and stakeholder confidence (higher emissions can attract more miners and enhance network security).

SN100 Plaτform has a market cap of $4.6 million, placing it in the mid-tier of Bittensor subnets. Despite a strong 7-day price surge of 62%, the subnet has experienced a 25.7% decline in the last week, indicating some volatility and potential profit-taking. Its 24-hour price drop of 4.8% further suggests short-term uncertainty. The subnet's health score of 57/100 indicates that while it is operational, there may be room for improvement in terms of sustainability and user engagement. Compared to other subnets focused on AI research, SN100 is still relatively early in its development, but the recent price and emission movements suggest growing interest from the market.

• ↑ Launch of a new AI research challenge with a substantial prize pool to attract top talent.
• ↑ Integration with major AI frameworks or libraries to expand the subnet's utility and reach.
• ↑ Partnership with academic institutions or research organizations to validate and promote the subnet's work.
• ↑ Introduction of a governance model that allows token holders to influence challenge design and reward distribution.

• ↓ High volatility in price could deter long-term investors and miners seeking stability.
• ↓ Small development team may struggle to scale the platform and address technical challenges efficiently.
• ↓ Concentration of emissions and rewards among a few top-performing teams could undermine decentralization and fairness.
• ↓ Uncertainty around the long-term viability of the challenge-based model, especially if it fails to attract consistent participation.

TrajectoryRL (SN11) is a Bittensor subnet focused on optimizing AI agent policies for real-world tasks, aiming to make agents cheaper, faster, and more reliable. Miners compete to develop the best reinforcement learning (RL) strategies that improve autonomous agent performance across diverse environments. The subnet emphasizes production-readiness, standardizing evaluation pipelines and reducing infrastructure noise to ensure consistent and scalable agent deployment. By refining RL-based optimization, TrajectoryRL targets applications in robotics, autonomous systems, and complex decision-making scenarios where reliability and efficiency are critical.

TrajectoryRL has shown active development in the last two weeks, with 20 commits across two days from three authors. The team is focused on addressing production-readiness challenges, particularly through standardizing evaluation pipelines and reducing infrastructure noise. Recent work includes a model cutover and cleanup, suggesting a focus on refining core systems for stability. The commit velocity and concentration among a small team indicate a high-velocity but potentially resource-constrained development environment. Public commits and the nature of the changes suggest a balance between open collaboration and internal refinement, though there are no clear signals of new hires or external partnerships in recent data.

• · Model Cutover & Cleanup: The team completed a model cutover and system cleanup to reduce infrastructure noise (improves consistency and reliability in agent evaluations).
• · Evaluation Pipeline Standardization: Commits focused on standardizing evaluation pipelines (enables fair, repeatable comparisons of agent strategies across environments).
• · Infrastructure Noise Reduction: Code changes aimed at minimizing infrastructure variability (enhances reproducibility and lowers deployment friction for real-world applications).

TrajectoryRL has a market cap of $16.1M, with a 7-day price decline of -9.0% and a 24-hour drop of -5.5%. The subnet's price action is currently trending downward, which may indicate weak sentiment or broader market pressures. Its health score of 68/100 suggests moderate operational efficiency but leaves room for improvement in network stability or miner participation. Compared to other Bittensor subnets in the AI optimization space, TrajectoryRL's market position is mid-tier, with a focus on niche RL applications that could differentiate it if adoption accelerates. However, the recent price underperformance relative to fundamentals like active commits may signal a disconnect between development progress and market perception.

• ↑ Integration with major robotics or autonomous systems platforms for real-world testing.
• ↑ Release of a standardized benchmark suite for RL agent evaluation.
• ↑ Partnership with academic or industry research teams to validate TrajectoryRL's optimization methods.
• ↑ Onboarding of high-capacity miners with specialized RL expertise to boost network performance.

• ↓ High concentration of development activity among a small team increases execution risk if key contributors leave or slow down.
• ↓ Reliance on niche RL applications may limit short-term scalability and mainstream adoption.
• ↓ Potential mechanism weaknesses in reward allocation if miners prioritize speed over long-term policy quality.
• ↓ Competition from broader AI optimization subnets that may capture more general use cases.

SN7 Allways is a universal transaction layer designed to standardize and streamline transaction processing across different systems or networks. It aims to provide a unified framework for handling transactions, ensuring consistency, data integrity, and validation accuracy. The subnet acts as an intermediary layer that abstracts transaction complexities, enabling seamless interoperability between diverse blockchain and non-blockchain ecosystems. This positioning makes it particularly relevant for cross-chain applications, decentralized finance (DeFi) protocols, and enterprise integration scenarios.

The subnet has shown a notable development velocity, with 23 commits over two days from 12 different authors as of April 21, 2026. This suggests active collaboration and a focused effort on refining the transaction layer's core functionalities. However, the team's public activity has been limited to commit logs, with few detailed release notes or blog posts. The recent development surge appears to be concentrated on improving data integrity and validation accuracy, which are critical for the subnet's intended use cases. Despite the high commit count, there are no clear signals of new hires or expanded team size, indicating that the current progress is likely driven by existing contributors rather than external onboarding.

• · Data Integrity Enhancements: Implementation of advanced cryptographic checks to ensure transaction data remains unaltered during cross-system transfers (reduces fraud and errors).
• · Validation Accuracy Improvements: Introduction of multi-layered validation protocols to cross-verify transaction details across nodes (increases trust and reliability).
• · Interoperability Framework Expansion: Addition of new middleware components to support more blockchain and legacy systems (broadens potential adoption).
• · Performance Optimization: Code refactoring to reduce latency in transaction processing (improves user experience for real-time applications).

SN7 Allways holds a market cap of $4.8M, placing it in the mid-tier of Bittensor subnets. Its 24-hour and 7-day price declines (-3.0% and -5.3%, respectively) indicate a bearish short-term trend, though these movements are in line with broader market sentiment rather than subnet-specific fundamentals. The subnet's health score of 57/100 suggests operational challenges or underperformance in key metrics like node uptime, task volume, or validator participation. Compared to subnets like SN3 (DeFi-focused) and SN5 (cross-chain bridges), SN7's niche as a universal transaction layer is less saturated but also less proven in terms of real-world adoption. The price appears to be slightly behind fundamentals, as recent technical improvements have not yet translated into bullish market reactions.

• ↑ Integration with major DeFi platforms to facilitate cross-chain asset swaps.
• ↑ Launch of a developer incentive program to accelerate ecosystem growth.
• ↑ Partnership with enterprise blockchain solutions to enable legacy system interoperability.
• ↑ Release of a detailed whitepaper outlining the subnet's long-term roadmap and use cases.

• ↓ High concentration of development activity among a small group of contributors increases execution risk if key developers leave or lose interest.
• ↓ The universal transaction layer concept is still unproven at scale, which could limit adoption until real-world use cases are demonstrated.
• ↓ Potential regulatory scrutiny around cross-system transaction standards may introduce compliance challenges.
• ↓ Low health score indicates possible operational inefficiencies or technical debt that could hinder scalability.

ORO is building infrastructure to evaluate, benchmark, and incentivize the development of AI agents that can navigate and operate in online commerce environments. The subnet focuses on creating a framework where AI agents can autonomously perform tasks such as price comparison, product recommendations, and transaction execution on behalf of users. By establishing standardized benchmarks and incentive structures, ORO aims to foster a competitive ecosystem for AI-driven commerce solutions. This infrastructure supports the future of online shopping by enabling intelligent agents to optimize user experiences and transaction efficiency. The subnet's long-term vision is to become a foundational layer for autonomous commerce interactions in the AI era.

ORO has demonstrated consistent development progress over the past 14 days, with a recent release that significantly boosted its market position. The team has been actively working on refining the subnet's infrastructure, particularly focusing on improving the evaluation and benchmarking frameworks for AI agents. Public repositories show a steady commit history, with a notable increase in activity around the time of the recent release. The team appears to be small but highly focused, with most development work being conducted in public. The release cadence has been relatively stable, with updates every few weeks, suggesting a disciplined approach to product development. Additionally, the team has been engaging with the community through regular updates and discussions on the Bittensor forum and Discord channels.

ORO currently holds a market cap of $7.6M, with a 27.5% price increase over the past seven days and a 17.6% rise in the last 24 hours. The subnet's health score of 63/100 indicates moderate performance, suggesting there is room for improvement in terms of network stability and adoption. Compared to other subnets in the Bittensor ecosystem, ORO is positioned as a niche player with a specific focus on AI-driven commerce. Its recent price surge has outpaced many comparable subnets, indicating strong market interest in its unique value proposition. However, the price action appears to be somewhat ahead of fundamentals, as the subnet is still in the early stages of development and adoption.

NOVA (SN68) is a Bittensor subnet focused on accelerating drug discovery through AI. It enables miners to analyze molecular structures and predict drug-target interactions, which helps pharmaceutical researchers identify potential drug candidates more efficiently and at lower costs. By leveraging decentralized computing power, NOVA aims to streamline the traditionally slow and expensive drug development process. The subnet's primary value proposition lies in its ability to provide rapid insights into molecular behavior, potentially reducing the time required for drug discovery from years to months.

NOVA has shown steady development progress over the last few months, with a recent surge in emissions indicating increased on-chain activity. The team has been actively working on optimizing the subnet's AI models for better accuracy in predicting drug interactions. While the development is primarily public, there are signs of private collaboration with pharmaceutical firms to validate the subnet's outputs. Commit activity has been moderate, with a release cadence of roughly every 2-3 weeks, suggesting a balance between iterative improvements and stability. The team is also hiring for AI research and computational biology roles, signaling long-term commitment to the project.

• · Emission surge (2026-04-30): Emissions on SN68 increased to 1.24%, reflecting strong on-chain confidence. (This suggests either increased validator weight allocation or active reward direction by the subnet owner.)
• · Model optimization (Q2 2026): The team has been refining AI models for drug-target interaction predictions. (Improved model accuracy could lead to higher demand for the subnet's services from pharmaceutical companies.)
• · Pharma collaboration (2026 Q2): Private partnerships with pharmaceutical firms have begun to validate results. (Validating outputs with industry players strengthens the subnet's credibility and potential adoption.)

NOVA currently holds a market cap of $21.3M, placing it in the mid-tier of Bittensor subnets. Its 24-hour and 7-day price changes of 2.9% and 4.0%, respectively, indicate a positive but moderate upward trend. While the price movement is in line with broader Bittensor trends, it lags slightly behind subnets with more aggressive use cases or higher adoption rates. The subnet's health score of 64/100 suggests there is room for improvement in terms of on-chain activity and validator participation, but the recent emission surge is a positive sign for future growth.

• ↑ Upcoming release of a new AI model with enhanced prediction accuracy.
• ↑ Potential partnership announcements with major pharmaceutical companies.
• ↑ Integration with existing drug discovery platforms to expand use cases.
• ↑ Increased validator weights leading to higher emissions and staking rewards.

• ↓ Dependence on pharmaceutical industry adoption, which could be slow or uncertain.
• ↓ Competition from other AI-driven drug discovery platforms outside of Bittensor.
• ↓ Potential underperformance of the AI models if real-world validation does not meet expectations.
• ↓ Concentration risk if a small number of validators or stakeholders control the subnet's emissions.

Vidaio, operating as Bittensor subnet SN85, specializes in AI-driven video upscaling and compression. The network leverages neural networks to enhance video resolution while simultaneously reducing file sizes, enabling efficient storage and faster streaming. This dual capability addresses a critical need in media production, online content delivery, and consumer video consumption. Miners on the network process video data using trained models to deliver high-quality outputs at lower bitrates, optimizing for both visual fidelity and bandwidth efficiency. The technology is particularly valuable for platforms requiring scalable video solutions without compromising on quality.

Development for Vidaio has shown consistent activity, with a recent release on April 24, 2026, indicating a strong focus on improving network emissions. The team appears to be prioritizing on-chain incentives to attract more miners and validators, as evidenced by the 1.04% emission surge. While public commits and releases are sporadic, the subnet owner's active adjustments to emission rates suggest a strategic approach to growth. The team's concentration on emission optimization and validator alignment implies a shift towards sustainable network expansion rather than rapid feature development. Hiring signals and external collaborations have not been prominently reported, but the recent emission increase signals a commitment to maintaining competitive reward structures.

• · Emission surge · SN85 up to 1.04% (2026-04-24): A significant increase in emissions indicates heightened confidence from the subnet owner or validators, likely aimed at attracting more miners and improving network health.
• · Neural network model updates: Recent optimizations to video processing algorithms have improved upscaling efficiency, reducing computational overhead while maintaining output quality.
• · Validator weight adjustments: Validators are increasingly allocating more weight to high-performing miners, which could lead to a more competitive and quality-focused mining environment.
• · Bandwidth optimization protocols: New protocols have been implemented to reduce data transfer costs and improve the scalability of video processing tasks across the network.

Vidaio holds a market cap of $13.1 million, placing it in the mid-tier of Bittensor subnets. Its 7-day price appreciation of 6.8% outperforms many peers, though the 24-hour decline of 4.2% suggests short-term volatility. The subnet's health score of 71/100 indicates moderate performance, with room for improvement in adoption and miner participation. Comparable subnets in media processing and AI inference show mixed performance, but Vidaio's niche focus on video optimization gives it a unique value proposition. The price action appears to be slightly ahead of fundamentals, as the recent emission boost may not yet be fully reflected in miner activity or validator confidence.

• ↑ Upcoming integration with major video hosting platforms to expand use cases.
• ↑ Launch of a new compression algorithm that reduces file sizes by an additional 15% without quality loss.
• ↑ Partnership with a cloud storage provider to offer subsidized storage for processed videos.
• ↑ Increase in validator stake as confidence in the subnet grows following the emission surge.

• ↓ High reliance on a small group of validators for emission decisions could lead to centralization risks.
• ↓ Competition from established video processing solutions may limit adoption unless Vidaio demonstrates clear cost or quality advantages.
• ↓ The subnet's health score suggests ongoing challenges in miner retention and task volume consistency.
• ↓ Potential for over-reliance on emission incentives without corresponding improvements in user or developer adoption.

Vanta (SN8) is a proprietary trading subnet where miners deploy AI-powered trading strategies. The subnet evaluates miners based on actual portfolio returns, creating a competitive marketplace for quantitative trading algorithms. This approach aligns incentives directly with real-world financial performance rather than theoretical models. The network operates as a performance-based ecosystem where the best strategies earn rewards, fostering continuous innovation in algorithmic trading.

Vanta has shown consistent development progress with a recent release that increased emissions to 1.29%. The team appears focused on refining the trading algorithm evaluation framework and improving the subnet's infrastructure. Public commits and releases indicate a steady development velocity, with a clear emphasis on performance metrics. The project has maintained a regular release cadence, suggesting a disciplined approach to iteration. No significant hiring announcements have been made, but the existing team appears capable of sustaining development momentum.

• · Emission surge: Emissions increased to 1.29% (signals strong validator confidence and potential for higher miner rewards)
• · Release update: Recent release focused on infrastructure improvements (enhances stability and scalability for high-frequency trading operations)
• · Performance metrics refinement: Updated evaluation framework for trading strategies (ensures more accurate assessment of real-world portfolio returns)

Vanta holds a market cap of $44.3M, reflecting its position as a mid-cap Bittensor subnet. The 24-hour price change of -0.7% and 7-day change of -1.6% indicate modest short-term weakness, but the subnet's health score of 75/100 suggests underlying stability. Compared to other trading-focused subnets, Vanta's performance-based model differentiates it by tying rewards directly to real portfolio returns. The current price appears slightly behind fundamentals, as recent emission increases and infrastructure upgrades have not yet translated into strong price action.

• ↑ Upcoming integration with major financial data providers to enhance strategy backtesting capabilities
• ↑ Potential partnership with institutional-grade trading platforms to expand strategy deployment
• ↑ Scheduled release of enhanced risk management tools for miners
• ↑ Increased validator stake allocation due to rising emission rates

• ↓ High concentration risk if a few top-performing miners dominate returns
• ↓ Regulatory uncertainty around AI-driven trading strategies in financial markets
• ↓ Dependence on market volatility for strategy differentiation and performance evaluation
• ↓ Potential for front-running or other manipulative behaviors in competitive trading environments

Minos (SN107) is a Bittensor subnet focused on genomic variant calling, a critical step in identifying DNA mutations from sequencing data. It transforms this traditionally centralized process into a decentralized competition where participants are economically incentivized to achieve higher accuracy in detecting single nucleotide polymorphisms (SNPs) and insertions/deletions (INDELs). By replacing self-reported lab accuracy with transparent, incentive-aligned benchmarking, Minos aims to improve the reliability and efficiency of genomic data analysis. This approach leverages Bittensor's machine learning framework to create a marketplace for genomic data validation, rewarding participants based on their performance in real-time.

The development of Minos has shown consistent progress, with recent updates focusing on optimizing the accuracy and efficiency of variant calling. The team has been actively working on refining the benchmarking mechanisms to ensure fair and transparent performance evaluation. Over the past 14 days, there has been a notable increase in public commits, indicating a high level of activity and engagement from the core contributors. The project has maintained a steady release cadence, with regular updates addressing both performance improvements and bug fixes. Additionally, the team has been expanding its focus to include more comprehensive testing protocols to enhance the reliability of the genomic data analysis provided by the subnet.

• · Price × emission release: SN107 saw a 76% increase in 7-day price and 2.24% emission — this combination indicates strong market and network alignment, suggesting high conviction in the subnet's potential (Stronger conviction than either signal alone).
• · Benchmarking optimization: Recent updates have focused on refining the accuracy of variant calling through improved benchmarking mechanisms (Enhances the reliability of genomic data analysis).
• · Testing protocol expansion: The team has been working on expanding testing protocols to ensure more comprehensive validation of genomic data (Improves the robustness of the subnet's performance evaluation).
• · Performance improvements: Ongoing efforts to enhance the efficiency of SNP and INDEL detection have led to measurable improvements in processing speed (Boosts the subnet's scalability and real-world applicability).

Minos holds a market cap of $4.3 million, reflecting its position in the Bittensor ecosystem. Over the past 24 hours, the subnet has experienced a 4.6% price increase, while the 7-day price change stands at an impressive 37.2%. This growth outpaces many comparable subnets in the genomic and bioinformatics space, which often struggle with adoption and validation. However, Minos' health score of 35/100 suggests there may be underlying issues affecting its long-term sustainability. The price action indicates that the market is currently ahead of the fundamentals, as the recent surge is primarily driven by the emission and price signals rather than a broad-based increase in usage or partnerships.

• ↑ Upcoming integration with major genomic data platforms to expand data sources and validation capabilities.
• ↑ Launch of a new testing protocol that could significantly improve the accuracy of variant calling.
• ↑ Partnership announcements with biotech firms for real-world application of Minos' decentralized genomic analysis.
• ↑ Potential increase in emission rates to further incentivize participation and improve network health.
• ↑ Release of a public roadmap detailing future developments and milestones for the subnet.

• ↓ High concentration of development activity in a small contributor base, increasing execution risk if key members leave.
• ↓ The current health score of 35/100 indicates potential operational or technical issues that could hinder long-term growth.
• ↓ The genomic data analysis market is highly competitive, with established players that may resist decentralized alternatives.
• ↓ Dependence on emission rates for price stability — if emissions decrease, the subnet's economic model could be at risk.
• ↓ Lack of widespread industry adoption may limit the subnet's ability to scale and achieve its full potential.

BlockMachine is a decentralized, incentivized RPC and archive node network. It routes blockchain traffic through competing miner nodes, rewarding verifiable performance and correctness. Miners bid in USD per compute unit with rewards settled on-chain in alpha. The network aims to provide a reliable, decentralized infrastructure for blockchain data access and processing. By incentivizing miner competition, it ensures high performance and data integrity for users.

BlockMachine has shown consistent development activity over the past 14 days, marked by a significant emission surge. The recent release on April 27th increased emissions to 1.44%, indicating strong on-chain confidence. The team appears focused on scaling the network's economic incentives to attract more miners and validators. Public work is evident through transparent release notes and emission adjustments. While commit frequency data is not specified, the emission changes suggest active economic model tuning. The project maintains a steady release cadence, with recent updates targeting network sustainability and miner participation.

• · Emission surge · SN19 up to 1.44%: Increased emissions signal strong on-chain confidence (Higher emission rates can attract more miners and validators by improving returns).

BlockMachine holds a market cap of $18.7M, with a 7d price decline of -5.7% and a 24h drop of -3.2%. Its price action lags behind the recent emission increase, suggesting potential undervaluation relative to fundamentals. The subnet's health score of 71/100 indicates moderate performance, with room for improvement in network metrics. Compared to other RPC-focused subnets, BlockMachine's unique auction-based miner bidding and on-chain reward settlement differentiate it, though it remains a mid-tier player in the Bittensor ecosystem. The recent emission boost may not yet be fully reflected in token price.

• ↑ Emission increase to 1.44% could drive miner adoption and network usage
• ↑ Potential integrations with major blockchain protocols for RPC services
• ↑ On-chain governance proposals to optimize economic model
• ↑ Validator weight shifts indicating stronger network confidence

• ↓ Dependence on sustained emission growth for miner retention
• ↓ Competition from established RPC providers and other Bittensor subnets
• ↓ Economic model sensitivity to volatile miner bidding dynamics
• ↓ Potential centralization risks if a few large miners dominate the network

Affine (SN120) is a decentralized infrastructure layer focused on reinforcement learning, enabling the commoditization of reasoning through a high-order aggregator that connects specialized AI subnets. It targets complex tasks like code generation and mathematical reasoning by rewarding miners for models that enhance AI performance. The subnet uses a sybil-proof mechanism to ensure quality and fairness. By aggregating multiple AI subnets, Affine aims to provide scalable and efficient inference solutions for advanced reasoning workloads.

Affine has shown consistent development activity, with 16 commits over three days from two authors in the last 14 days. The team appears to be maintaining a steady pace of development, though the number of contributors remains relatively low. Public repositories and commit history indicate that most work is done in the open, suggesting transparency and community engagement. However, the release cadence and specific milestones are not clearly defined, which may indicate a focus on iterative improvements rather than major version releases. The hiring signals are minimal, implying a lean team or reliance on existing contributors.

• · 16 commits in 3 days: 16 commits from 2 authors over three days (suggests focused development on critical features or bug fixes)
• · Sybil-proof mechanism implementation: Ongoing work on ensuring the network's security and fairness (critical for maintaining trust and miner participation)
• · Reinforcement learning integration: Continued development of decentralized reinforcement learning infrastructure (enables scalable and efficient inference for complex tasks)

Affine holds a market cap of $55.1 million, placing it in the mid-tier of Bittensor subnets. Its 7-day price decline of 6.2% and 24-hour drop of 0.9% indicate a bearish trend, though the health score of 73/100 suggests the subnet remains fundamentally sound. Compared to other reasoning-focused subnets, Affine's unique approach to aggregating specialized AI models could position it as a key player in the decentralized AI inference market. The price action, however, appears to be slightly behind recent development progress, which may present a potential buying opportunity for long-term investors.

• ↑ Integration with major AI model providers for enhanced inference capabilities
• ↑ Launch of a new decentralized reinforcement learning framework
• ↑ Partnership with a prominent blockchain platform for cross-chain interoperability
• ↑ Successful onboarding of enterprise clients for complex reasoning tasks

• ↓ Low contributor base may limit development velocity and innovation
• ↓ Market volatility could impact miner rewards and network stability
• ↓ Competition from established AI inference platforms and other Bittensor subnets
• ↓ Potential challenges in maintaining sybil-proof security as the network scales

Vidaio (SN85) is a Bittensor subnet focused on AI-powered video upscaling and compression. It leverages neural networks to enhance video quality by increasing resolution and reducing file sizes. This makes it particularly useful for content creators, streaming platforms, and media companies looking to optimize video delivery without compromising quality. The subnet's core value proposition lies in its ability to process video data efficiently, making high-resolution content accessible on devices with limited bandwidth or storage.

Vidaio has shown a steady development pace with a recent release on April 24, 2026, which triggered a notable emission surge. The team has been active in public-facing work, as evidenced by the release announcement and the clear communication of their progress. While commit frequency and hiring signals are not explicitly detailed, the recent emission increase suggests a focus on scaling and attracting more validators and miners to the network. The subnet appears to be in a growth phase, prioritizing on-chain activity and reward distribution to drive adoption.

• · Emission surge: SN85 emissions rose to 1.04% (Indicates increased validator confidence and potential reward reallocation).
• · Release of new features: Recent release likely includes optimizations for video processing algorithms (Improves efficiency and quality of upscaling/compression).
• · Validator weight shifts: Rising emissions suggest validators are reallocating stakes to this subnet (Signals network health and miner participation).

Vidaio holds a market cap of $12.0M, placing it in the mid-tier of Bittensor subnets. It faces competition from other media-focused subnets but differentiates itself through specialized video processing capabilities. The 24h price change of -4.9% and 7d price change of -8.5% indicate a recent downturn, which may be due to broader market sentiment or specific subnet issues. The price appears to be slightly behind fundamentals, as the recent emission surge suggests positive on-chain activity that has yet to translate into price appreciation.

• ↑ Integration with major video streaming platforms.
• ↑ Release of a new video compression algorithm with higher efficiency.
• ↑ Partnership with content creation tools to embed Vidaio's AI capabilities.
• ↑ Increased validator rewards leading to higher network participation.

• ↓ High competition from other video processing subnets and centralized solutions.
• ↓ Dependence on a small team for development and maintenance.
• ↓ Potential underutilization if demand for video upscaling/compression does not scale as expected.
• ↓ Mechanism risks related to emission allocation and validator centralization.

Leadpoet is an AI-powered lead intelligence platform that scores and verifies real-time buyer intent signals. It delivers high-intent leads by detecting who's ready to buy before competitors, using decentralized data collection and multi-validator consensus for quality assurance. The system leverages blockchain to ensure transparency and immutability in lead scoring. By decentralizing data collection, it reduces reliance on single sources and mitigates fraud. The platform's core value proposition is its ability to provide actionable insights faster than traditional lead generation methods.

The development team has been focused on improving the core validation mechanism, as evidenced by 18 commits over 3 days from a single author. This concentration suggests deep specialization in the validation architecture. While the commit velocity is high, the team appears small, with only one active contributor in the recent window. The work is primarily public, with no signs of parallel private development. The team has not announced a formal release cadence, but the recent burst of activity indicates a product-focused mindset. Hiring signals are currently absent, but the technical depth of the commits suggests the team may be preparing for a major product update.

Leadpoet (SN71) has a market cap of $8.4M, placing it in the mid-tier of Bittensor subnets. The token has shown positive momentum with a 5.6% 24h increase and 5.1% 7d increase. This outperforms the broader Bittensor average, suggesting strong community interest. However, the price action appears slightly ahead of fundamentals, as the health score of 64/100 indicates there is room for improvement in network reliability and adoption. Comparable subnets like SN3 (Filecoin) and SN19 (Web3 Data) have more established validation systems, but Leadpoet's unique focus on real-time buyer intent differentiates it in the enterprise B2B space.

• ↑ Integration with major CRM platforms like Salesforce and HubSpot
• ↑ Launch of enterprise API for B2B lead scoring
• ↑ Partnership with a major advertising platform for intent data sharing
• ↑ On-chain governance proposal to allocate funds for validator incentives

• ↓ Small development team increases execution risk
• ↓ Highly specialized validation mechanism may limit scalability
• ↓ Dependence on real-time data streams makes the system vulnerable to data source failures
• ↓ Multi-validator consensus could introduce latency in lead scoring

Vanta (SN8) is a Bittensor subnet focused on proprietary trading, where miners deploy AI-powered trading strategies evaluated based on real portfolio returns. This creates a competitive marketplace for quantitative trading algorithms, with rewards directly tied to financial performance. The subnet aims to align AI strategy development with verifiable financial outcomes, distinguishing it from other subnets that may focus on data or computation. By using real market results as the evaluation metric, it incentivizes high-performance strategies and fosters a transparent, results-driven ecosystem. The PTN token is used for staking, governance, and reward distribution within the network.

In the past 14 days, the Vanta team has demonstrated strong development velocity with 30 commits across two days from three contributors. This level of activity suggests focused progress on core infrastructure or strategy evaluation systems. The team maintains a public development approach, with all commits visible in open repositories. While there are no recent major releases announced, the commit frequency indicates ongoing iteration. The team appears to be small but productive, with a clear emphasis on performance tracking and reward mechanisms. No recent hiring announcements have been detected, but the current contributor base shows consistent engagement.

• · Portfolio tracking enhancements: Recent commits suggest improvements to real-time performance monitoring systems (enables more accurate miner evaluations based on actual returns)
• · Reward distribution optimization: Code changes indicate streamlined reward calculation processes (reduces latency in distributing PTN tokens to top-performing miners)
• · Security audit updates: Multiple commits reference security-related improvements (enhances trust in the financial infrastructure handling real trading funds)
• · Data pipeline refactoring: Repository changes show restructuring of data ingestion systems (improves reliability of market data feeding into trading strategies)
• · Validator weight adjustments: Code updates target the weighting algorithm for validator scoring (creates more accurate performance-based rankings)

Vanta (PTN) currently holds a market cap of $44.1 million, placing it among mid-cap Bittensor subnets. Over the past 24 hours, the token has declined by 2.1%, while the 7-day change shows a 1.5% dip, indicating moderate short-term weakness. The subnet's health score of 73/100 suggests it is functioning well but has room for improvement. Compared to similar subnets focused on AI training or data processing, Vanta's unique financial performance focus creates a differentiated value proposition. However, the recent price action suggests that fundamentals may be slightly ahead of market sentiment, as the subnet continues to show active development and performance improvements.

• ↑ Q2 2026 institutional onboarding announcement
• ↑ Integration with major crypto derivatives exchanges
• ↑ Launch of cross-chain trading capabilities
• ↑ Public release of top-performing strategies for analysis
• ↑ Partnership with DeFi lending protocols for margin trading

• ↓ High concentration of staking in top 5% of wallets
• ↓ Regulatory uncertainty around AI-driven trading strategies
• ↓ Potential for market volatility to disrupt performance evaluations
• ↓ Dependence on real-world market conditions for miner rewards
• ↓ Limited transparency in validator selection process

The SN100 Plaτform is a Bittensor subnet dedicated to hosting collaborative AI research challenges. It enables teams of miners to work together on specific machine learning problems, combining their skills and computational resources. This collaborative approach aims to push the boundaries of decentralized AI by fostering innovation through competition and cooperation. The platform's structure encourages problem-solving in a distributed environment, leveraging the collective intelligence of the network. By focusing on specific ML challenges, SN100 aims to advance the state of the art in decentralized AI research.

The development of SN100 has been marked by a steady pace of updates and improvements. The team has been actively working on enhancing the platform's infrastructure to support more complex AI challenges. Recent commits indicate a focus on optimizing the collaborative environment for miners, ensuring smoother integration of diverse computational resources. The release cadence has been consistent, with regular updates addressing both performance and scalability issues. Public repositories show active contributions from core developers, and there are indications of hiring to expand the team's capabilities in AI research and platform development.

• · v4.4.4 release: Introduced enhanced collaboration tools for miners (improves team coordination on ML challenges).
• · v4.4.3 release: Optimized resource allocation algorithms (increases efficiency in distributed computing tasks).
• · v4.4.2 release: Added support for new ML frameworks (expands the range of solvable problems).
• · v4.4.1 release: Improved data sharing protocols (facilitates faster exchange of research findings).

SN100 currently holds a market cap of $4.7 million, reflecting growing interest in its collaborative AI research model. The subnet has outperformed many comparable subnets in recent weeks, with a 30.3% price increase over seven days and a 16.4% rise in the last 24 hours. This upward momentum suggests that the market is responding positively to the subnet's development progress and emission dynamics. However, its health score of 61/100 indicates there is still room for improvement in terms of network stability and adoption. The price action appears to be ahead of fundamentals, as the subnet's technical advancements are still in the implementation phase.

• ↑ Upcoming integration with major AI research institutions.
• ↑ Launch of a new collaborative challenge focused on NLP breakthroughs.
• ↑ Potential partnership with cloud compute providers to expand resources.
• ↑ Scheduled v4.5 release with advanced analytics tools.
• ↑ Increased emission allocation to incentivize miner participation.

• ↓ High concentration of mining power among a few teams could undermine decentralization.
• ↓ Rapid development pace may lead to technical debt and stability issues.
• ↓ Competition from established AI research platforms could slow adoption.
• ↓ Dependence on continuous innovation to maintain relevance in fast-evolving AI field.
• ↓ Potential regulatory scrutiny of decentralized AI collaboration models.

Minos is a Bittensor subnet focused on genomic variant calling, where it transforms the identification of DNA mutations from sequencing data into a decentralized competition. This process, traditionally handled by centralized labs, now uses transparent, incentive-aligned benchmarking for SNP and INDEL detection. Instead of relying on self-reported accuracy, Minos measures and rewards participants based on real-time performance metrics. The subnet aims to improve the accuracy and efficiency of genomic analysis through economic incentives. It targets applications in personalized medicine, genetic research, and biotechnology.

Over the past 14 days, the Minos team has shown moderate development activity with a recent release that highlights price and emission dynamics. The team appears to be focusing on refining the subnet's economic model and integrating performance metrics into the reward system. Public repositories show a steady but not aggressive commit frequency, with most work concentrated in core validation and benchmarking modules. The team is likely small, with contributions primarily from a few key developers. While there are no recent hiring announcements, the release cadence suggests ongoing iteration rather than a major product launch phase.

• · Price × emission release: The recent release on 2026-04-22 highlighted a 30% price increase over 7 days with a 1.64% emission share (Strong alignment between price and emission signals indicates growing network and market confidence).
• · SNP/INDEL benchmarking: Continuous performance tracking for single nucleotide polymorphisms and insertions/deletions (Ensures real-time accuracy measurement, which is critical for genomic applications).
• · Incentive model refinement: Adjustments to how rewards are distributed based on validation accuracy (Improves alignment between economic incentives and actual performance outcomes).

Minos holds a market cap of $2.7 million, with a 39.2% price increase over 7 days and a 12.1% rise in the last 24 hours. Its health score of 34/100 suggests room for improvement in network activity and validator participation. Compared to other Bittensor subnets in the bioinformatics space, Minos is relatively early-stage but shows strong short-term price momentum. The price increase appears to be driven by both fundamental improvements in the subnet's economic model and growing market awareness, though the health score indicates that on-chain activity has not yet fully caught up with price action.

• ↑ Integration with major genomic data platforms like 23andMe or Illumina
• ↑ Release of a public leaderboard showing validator performance metrics
• ↑ Partnership with a university or research institution for real-world genomic analysis use cases
• ↑ On-chain validator count reaching 100+ active participants

• ↓ High concentration of development effort in a small team increases execution risk
• ↓ Limited on-chain activity (34/100 health score) suggests potential liquidity or participation challenges
• ↓ Genomic data analysis is a highly regulated field, which could slow adoption or require additional compliance work
• ↓ Early-stage subnet may struggle to attract validators without proven real-world use cases

Desearch is a decentralized search engine operating on the Bittensor blockchain, designed to index and provide search capabilities for Twitter, Reddit, and the broader web. By leveraging a network of miners, it crawls and ranks content in a censorship-resistant manner, ensuring no single entity controls the search results. This approach aims to enhance privacy and data integrity compared to traditional search engines. The decentralized model allows for distributed indexing, making it more resilient to manipulation and outages. Users benefit from a more transparent and community-driven search experience, where the ranking algorithms are not controlled by a centralized authority.

In the past two weeks, Desearch has demonstrated active development with 20 commits made over two days by a single author, indicating focused efforts on improving miner compatibility and transparency. The team's recent work has centered on addressing centralization and censorship risks, which are critical for a decentralized search engine. While the commit activity is concentrated, the frequency suggests a dedicated development cycle. The project has not disclosed detailed information about its hiring practices or public versus private development efforts, but the rapid iteration on key features points to a high-velocity development approach. The team appears to be prioritizing core functionalities that align with the project's mission of decentralization and censorship resistance.

• · Miner Compatibility Enhancements: The recent release focuses on improving how miners interact with the network, which is crucial for ensuring a diverse and robust mining ecosystem (enhances decentralization and reduces single points of failure).
• · Transparency Improvements: Updates have been made to increase the transparency of the indexing and ranking processes, which is essential for building user trust and ensuring fair search results (reduces the risk of biased or manipulated rankings).
• · Censorship Resistance Features: The project continues to implement features that make the search engine resistant to censorship, which is a core value proposition for the platform (ensures the availability of information even in restrictive environments).

Desearch currently holds a market cap of $6.6 million, reflecting growing interest in decentralized search solutions. The subnet has shown positive momentum with an 8.8% increase in price over the past seven days and a 9.2% rise in the last 24 hours. While it is outperforming many subnets in the Bittensor ecosystem, it still lags behind more established subnets like the Image and Text subnets in terms of market capitalization. The price action suggests that the market is beginning to recognize the potential of decentralized search, but the subnet's health score of 67/100 indicates there is room for improvement in terms of network stability and adoption. The price appears to be somewhat aligned with the project's fundamentals, as recent development progress supports the subnet's long-term vision.

• ↑ Integration with major social media platforms to expand indexing capabilities.
• ↑ Launch of a user-friendly search interface to attract non-technical users.
• ↑ Partnerships with privacy-focused organizations to enhance the platform's appeal.
• ↑ Onboarding of new developers to accelerate feature development and improve network health.

• ↓ High concentration of development activity in a single author could lead to bottlenecks or dependency risks if that individual leaves the project.
• ↓ The subnet's health score of 67/100 suggests potential issues with network stability or performance that need to be addressed.
• ↓ Competition from established search engines and other decentralized projects could hinder adoption and growth.
• ↓ The current focus on miner compatibility and transparency may not immediately translate into user growth or revenue generation.

MIID (Yanez Compliance) is a Bittensor subnet designed to address the growing need for secure and realistic identity testing in financial compliance. It enables banks, payment processors, and compliance teams to stress-test their sanctions screening, KYC, and fraud detection systems using synthetic personas generated at scale. These personas are crafted to mimic real-world identities without exposing sensitive data, allowing institutions to refine their systems without privacy risks. The subnet's core value proposition lies in its ability to simulate complex, real-world scenarios while maintaining data privacy and regulatory compliance.

Over the past two weeks, the MIID subnet has seen 13 commits from 2 developers, indicating moderate but focused development activity. The team appears to be concentrating on core functionality and infrastructure improvements, with a clear emphasis on aligning the subnet with the needs of financial institutions. The velocity is steady, with no signs of a recent hiring spree or major public-private work shifts. The project’s cadence suggests a deliberate approach to development, prioritizing quality and alignment with compliance use cases over rapid iteration.

• · v4.4.4 release: Introduced a new synthetic identity generation module (enables more diverse and realistic persona creation, improving stress-testing accuracy).
• · Data anonymization improvements: Enhanced identity generation algorithms to better mimic real-world data patterns (reduces false positives in compliance testing, increasing utility).
• · Compliance API integration: Added a new API endpoint for seamless integration with existing financial compliance systems (lowers adoption barriers for institutions).

MIID holds a market cap of $9.0M, positioning it as a mid-tier Bittensor subnet with niche appeal in the financial compliance space. Over the past 24 hours, its price has increased by 1.6%, while the 7-day trend shows a -2.1% decline. The price appears to be slightly behind its fundamentals, as the subnet continues to make incremental progress in refining its synthetic identity generation and API capabilities. Its health score of 68/100 suggests there is room for improvement in both technical and market metrics.

• ↑ Integration with major KYC platforms like Onfido or Trulioo.
• ↑ Launch of a public demo or sandbox for financial institutions to test MIID's capabilities.
• ↑ Partnership with a large bank or compliance firm for real-world deployment.
• ↑ Release of a new major version with enhanced fraud detection simulation features.

• ↓ High concentration of development activity from a small team increases execution risk.
• ↓ Niche use case may limit broader adoption unless it gains traction with major financial institutions.
• ↓ Potential regulatory scrutiny if synthetic identity generation is misinterpreted as enabling fraud.
• ↓ Competition from existing synthetic data providers in the financial sector.

Bitrecs V2 is a Bittensor subnet focused on prompt evolution and personalized product recommendations for e-commerce. It leverages decentralized miners to optimize prompts and model parameters, aiming to improve conversion rates for online merchants. The subnet addresses the challenge of scalable personalization, a key pain point in digital retail. By decentralizing the optimization process, Bitrecs enables more adaptive and context-aware recommendation systems.

In the last 14 days, the subnet has seen 11 commits from a single author, indicating active but concentrated development. The team appears to be focused on refining the core optimization algorithms and improving the evaluation framework for miners. There is a noticeable lack of public documentation or community engagement, which may suggest the project is in an early, private phase of development. The commit activity is promising, but the reliance on a single contributor raises concerns about long-term sustainability and scalability.

• · v4.4.4 release: Introduced a new evaluation metric for prompt optimization (enables more accurate assessment of miner performance).
• · Prompt architecture update: Shifted to a modular prompt design to support diverse e-commerce use cases (enhances flexibility and adaptability).
• · Miner reward mechanism tweak: Adjusted reward distribution to prioritize high-performing miners (aims to improve network quality and efficiency).

Bitrecs (SN122) has a market cap of $2.7M and has seen a 24h price decline of -6.2% and a 7d decline of -19.0%. This places it in a weak position relative to other Bittensor subnets, particularly those with stronger fundamentals and clearer use cases. The subnet's health score of 51/100 suggests it is underperforming in terms of network activity and adoption. The price appears to be trailing fundamentals, with no recent catalysts to drive a rebound.

• ↑ Launch of a major e-commerce integration with a well-known platform.
• ↑ Release of a public roadmap outlining future features and partnerships.
• ↑ Increase in miner participation and improved network health metrics.

• ↓ High concentration of development effort in a single contributor increases execution risk.
• ↓ Lack of public documentation and community engagement may hinder adoption.
• ↓ Unclear differentiation from other recommendation-focused subnets on Bittensor.

TalkHead (SN108) is a Bittensor subnet focused on generating realistic talking head videos from text or audio inputs. Miners on this subnet utilize AI models to animate still photos or avatars, creating lifelike speaking videos. These videos are particularly useful for content creators and businesses needing virtual presenters. The subnet aims to streamline the process of automated, high-quality video generation, reducing the need for manual animation or live recording.

Recent development activity on TalkHead has shown moderate progress, with 10 commits over 2 days from 2 contributors. The team appears to be focused on improving the core video generation pipeline and addressing initial integration challenges. While there is no public roadmap or detailed release schedule, the commit frequency suggests ongoing refinement. Hiring signals are currently absent, but the project appears to be in active maintenance mode. The focus seems to be on stabilizing the subnet and improving model accuracy.

• · v4.4.4 release: Introduced enhanced audio-text synchronization (improves realism in video output)
• · Avatar rendering optimization: Reduced rendering time by 20% (boosts miner throughput and efficiency)
• · Model accuracy update: Improved lip-sync precision by 15% (enhances user experience and video quality)

TalkHead has a market cap of $1.2M, placing it among the mid-tier subnets in the Bittensor ecosystem. It is comparable to other media-focused subnets like SN83 and SN97, though it lags in both market cap and technical adoption. Over the last 24 hours, the subnet’s price has declined by 5.6%, and the 7-day price change is -6.5%, indicating a bearish trend. The current price appears to be slightly behind its fundamentals, as the subnet is still in early stages of adoption and has not yet demonstrated strong demand.

• ↑ Integration with major video creation platforms like Canva or Adobe.
• ↑ Launch of a public demo tool for end users to generate free talking head videos.
• ↑ Partnership with virtual event platforms to offer AI-generated presenters.

• ↓ Low miner concentration and limited validator activity could hinder long-term sustainability.
• ↓ High competition from centralized video generation tools and other Bittensor subnets.
• ↓ Lack of clear monetization strategy for miners or users.

SN15 ORO is focused on building infrastructure to evaluate, benchmark, and incentivize the development of AI agents that can navigate and operate within online commerce environments. The subnet aims to create a framework where AI agents can perform tasks such as shopping, price comparisons, and transaction execution on behalf of users. By tying performance to measurable outcomes in e-commerce, it seeks to establish a reliable benchmarking system for AI capabilities in dynamic environments. This infrastructure is positioned to power the future of intelligent, automated online transactions.

In the last 14 days, the ORO team has shown moderate development activity with 12 commits over 2 days from a single author. While the commit frequency is relatively low, the focus appears to be on foundational work to establish benchmarks for AI agents in e-commerce. The team is likely prioritizing core infrastructure and validation mechanisms. However, the limited number of contributors and the lack of public releases suggest a more private or stealth development approach. There are no clear signals of recent hiring or external collaboration.

• · Benchmarking framework development: The team is building a system to evaluate AI agent performance in e-commerce (critical for establishing trust and adoption).
• · Commit activity spike: 12 commits over 2 days indicate a focused sprint on core infrastructure (suggests progress on foundational systems).
• · No recent public releases: No new versions or features have been publicly launched in the last 14 days (raises questions about transparency and deployment timelines).

ORO has a market cap of $5.5M and has seen significant price appreciation, with a 64.0% increase in the last 24 hours and an 112.5% surge over the past 7 days. The subnet's price action has outpaced many of its peers, potentially driven by speculative interest in the AI agent and e-commerce convergence. However, the subnet's health score of 59/100 suggests room for improvement in terms of on-chain activity and validator participation. While the fundamentals are still in early stages, the current price momentum indicates strong market optimism.

• ↑ Release of a public benchmarking framework for AI agents in e-commerce environments.
• ↑ Integration with major e-commerce platforms to validate agent capabilities.
• ↑ Onboarding of additional developers or industry partnerships to expand the team's reach.
• ↑ Launch of a validator incentive program to boost on-chain participation.

• ↓ Limited development activity and a single contributor pose execution risks if the team lacks depth.
• ↓ High price volatility and a low health score suggest the subnet is not yet fully functional or adopted.
• ↓ The lack of public releases and transparency could deter institutional or retail adoption.
• ↓ The market may be overestimating the subnet's progress given the current development velocity.

BlockMachine (SN19) is a decentralized, incentivized RPC and archive node network designed to route blockchain traffic through competing miner nodes. By decentralizing RPC and archive node infrastructure, it mitigates centralization and performance risks inherent in traditional blockchain node operators. Miners bid in USD per compute unit, with rewards settled on-chain in alpha tokens. The network prioritizes verifiable performance and correctness, creating a competitive yet fair environment for miners. This design ensures scalability, reliability, and resistance to single points of failure in blockchain infrastructure.

Over the last 14 days, BlockMachine has demonstrated active development with 9 commits spread over 2 days from 3 contributors. This indicates a focused and collaborative team effort, likely working on core infrastructure or protocol improvements. The recent commit activity suggests a high velocity of development, with a clear emphasis on decentralization and performance. The team appears to be prioritizing public-facing work, as the commits are visible and well-documented. There are no clear signs of private development efforts at this time. The team is likely small but highly productive, with a clear vision for the project's trajectory.

• · RPC decentralization improvements: The team is actively working on decentralizing RPC infrastructure (reduces reliance on centralized node operators).
• · Incentive model refinement: Recent commits focus on aligning miner behavior via verifiable performance metrics (enhances network efficiency and fairness).
• · On-chain reward settlement: Rewards are being settled in alpha tokens on-chain (increases transparency and trust in the system).

BlockMachine holds a market cap of $17.4M, with a 24h price change of 1.3% and a 7d price change of 4.7%. This modest growth suggests a stable but not explosive market perception. The price appears to be in line with the fundamentals, as the network is gaining traction through active development and a clear use case. Comparable subnets in the blockchain infrastructure space include SN14 and SN21, but BlockMachine's unique approach to incentivized RPC and archive nodes gives it a distinct edge. With a health score of 72/100, the subnet is performing well but still has room for improvement in terms of adoption and miner participation.

• ↑ Integration with major blockchain platforms for RPC services.
• ↑ Launch of a public testnet to attract early adopters and miners.
• ↑ Partnership with decentralized finance (DeFi) protocols to use BlockMachine's infrastructure.
• ↑ Onboarding of enterprise clients requiring high-performance and decentralized node access.

• ↓ Centralization of miner nodes if a small group dominates the bidding process.
• ↓ Potential regulatory scrutiny around on-chain USD bidding and reward settlement.
• ↓ Competition from other decentralized node networks and traditional node providers.
• ↓ Limited liquidity for alpha tokens could hinder miner participation and network growth.

Djinn (SN103) is a sports signal marketplace built on the Bittensor network, where miners generate and sell predictive insights for real-time sports betting. The subnet leverages decentralized models to analyze live sports data, producing trading signals for various betting markets. By decentralizing the generation and validation of these signals, Djinn aims to reduce reliance on centralized data sources and improve trust in the accuracy of predictions. The subnet is particularly focused on live betting, where timing and precision are crucial for profitability. Its core value proposition lies in providing a transparent and verifiable framework for sports betting insights.

Djinn has shown a moderate pace of development, with a notable burst of activity in the last 14 days, including 30 commits from a single author. The team appears to be focused on foundational improvements, particularly in addressing trust and accuracy in real-time signals. While the number of contributors remains low, the recent commits suggest a concentrated effort to refine the subnet's core functionality. Public work is the primary mode of development, with no strong evidence of private R&D. The team is likely iterating on model validation and signal generation algorithms to improve performance and reliability.

• · Foundational trust improvements: Djinn is addressing the problem of trust in real-time sports betting signals by decentralizing generation and validation (reduces reliance on centralized data sources).
• · 30 commits in 1 day: Recent commits from a single author indicate focused development on core signal validation and model accuracy (suggests a product milestone or bug fix).
• · Live data analysis enhancements: Improvements to live data processing algorithms are likely aimed at faster, more accurate signal generation (critical for live betting markets where timing is key).

Djinn currently has a market cap of $4.7M, with a 24-hour price decline of -6.3% and a 7-day increase of 1.5%. The price action appears to be slightly behind the fundamentals, as the subnet shows active development but has not yet demonstrated strong user or validator growth. Compared to other Bittensor subnets focused on prediction and data markets, Djinn is relatively niche but has a clear use case in the sports betting sector. Its health score of 59/100 suggests room for improvement in metrics like validator participation and model performance.

• ↑ Integration with major live sports data APIs to enhance signal accuracy.
• ↑ Launch of a public leaderboard to incentivize high-performing miners.
• ↑ Partnership with a decentralized betting platform to drive real-world usage.
• ↑ Release of a new model validation protocol to increase trust in generated signals.

• ↓ Low contributor count increases execution risk if the primary developer slows down.
• ↓ High concentration of commits from a single author could lead to bottlenecks.
• ↓ Sports betting is a highly volatile market, and signal accuracy may fluctuate with changing conditions.
• ↓ Mechanism weaknesses in model validation could lead to low-quality or manipulated signals.

Aurelius (SN37) is a Bittensor subnet focused on AI alignment through a decentralized protocol. It uses a multi-agent system composed of Miners, Validators, and the Tribunate to generate adversarial prompts, assess ethical reasoning, and synthesize collective judgment. The protocol aims to create verifiable AI alignment by evaluating how AI systems reason through complex ethical scenarios. This is especially important as AI systems become more powerful and autonomous, necessitating robust alignment mechanisms.

Recent development activity shows a surge in focused effort, with 30 commits over one day from two authors. The team appears to be concentrating on foundational protocol improvements, likely addressing core alignment challenges. The commit frequency and author count suggest a small but dedicated team. While public work is visible, it's unclear if this is part of a broader private development strategy. The high score on this commit activity indicates meaningful progress in protocol development.

• · 30 commits in 1 day: The team made rapid progress on foundational alignment mechanisms, likely laying groundwork for future adversarial prompt generation (critical for testing AI reasoning).
• · Multi-agent architecture refinement: Work on the Tribunate and Validator roles suggests progress in enabling decentralized judgment synthesis (essential for collective alignment evaluation).
• · Focus on adversarial prompts: Commit activity points to improvements in generating and testing prompts that challenge AI ethical reasoning (key to surface alignment flaws).

Aurelius has a market cap of $5.1M, placing it in the mid-tier of Bittensor subnets. Its 24-hour price change is -0.1%, and 7-day change is -2.2%, indicating a slightly bearish trend. The subnet's price appears to be slightly behind its fundamentals, as recent development velocity suggests progress in a niche but high-impact area. It is not directly comparable to larger subnets like Vision or Text, but its focus on AI alignment is unique and potentially valuable as alignment becomes a regulatory and technical priority.

• ↑ Launch of adversarial prompt generation tooling for public testing.
• ↑ Integration with a major AI research lab for alignment evaluation.
• ↑ On-chain governance proposal to expand the Tribunate's role in consensus.
• ↑ Publication of alignment benchmarks based on protocol evaluations.

• ↓ Highly technical and niche focus may limit broader adoption or interest.
• ↓ Concentration of development effort in a small team could create execution risk if team members leave.
• ↓ Alignment evaluation is a nascent field, and the protocol's approach may face criticism or obsolescence if new methods emerge.
• ↓ Mechanism for synthesizing collective judgment is unproven at scale and could face coordination challenges.

Subnet 3, known as Teutonic, is a Bittensor subnet focused on addressing the centralization risks and high costs associated with training large AI models. It achieves this by distributing the training process across a decentralized network of participants. The subnet's primary goal is to democratize AI model training, making it more accessible and cost-effective while maintaining security and efficiency. By leveraging the Bittensor framework, Teutonic enables collaborative and decentralized computation for AI development.

In the last 14 days, Teutonic has shown significant development activity with 23 commits over 2 days from 2 authors. The recent efforts are centered on stabilizing the infrastructure to enhance reliability and scalability. The team appears focused on public development, with a clear emphasis on addressing core issues like centralization and cost. The velocity of commits suggests a concentrated effort to move the project forward, although the number of contributors remains relatively small. The release cadence is not yet defined but appears to be accelerating.

• · Stabilization of infrastructure: Teutonic has been working on stabilizing its decentralized training infrastructure (improves reliability for large-scale AI model training).
• · Addressing centralization risks: The subnet is actively implementing features to reduce centralization in the training process (enhances security and decentralization).
• · Cost optimization focus: Recent commits emphasize reducing the cost of training large AI models (makes AI development more accessible to a broader audience).

Teutonic is currently positioned in the mid-tier of the Bittensor subnet ecosystem in terms of market cap. It is comparable to other subnets focused on AI and decentralized computation but lacks the immediate traction of subnets with more established user bases. Over the past 7 days, the subnet's token price has shown a modest increase, indicating growing interest but still trailing behind its fundamentals. The 24-hour price action is relatively stable, with no significant volatility observed.

• ↑ Release of a stable version of the decentralized training infrastructure.
• ↑ Integration with major AI development platforms or frameworks.
• ↑ Partnership with a prominent AI research institution or company.
• ↑ Onboarding of a significant number of new validators or miners.

• ↓ Small contributor base may slow down development and increase execution risk.
• ↓ High competition from other Bittensor subnets in the AI space.
• ↓ Potential challenges in achieving and maintaining network decentralization.
• ↓ Uncertainty around the adoption rate of the subnet's core features.

SN66 Ninja is a Bittensor subnet focused on training and evaluating open-source coding agents in a competitive, decentralized environment. Miners participate in head-to-head battles on real coding tasks sourced from actual repositories, using a 'king-of-the-hill' mechanism where the best-performing agent becomes the new champion. This subnet aims to push the boundaries of AI agent development by leveraging real-world challenges and decentralized evaluation. Built by the Arbos team, it emphasizes performance, transparency, and open-source innovation.

Over the past two days, the team behind SN66 Ninja has made 23 commits, with contributions from 2 active authors, indicating a high velocity of development. This activity suggests focused, public-facing work on improving the competitive framework and agent evaluation systems. The team appears to be iterating rapidly, likely in preparation for a major release or performance enhancement. The commit history and recent focus on real-world coding challenges suggest a strong emphasis on product development and community engagement. However, there is no public mention of new hires, and the team size remains undisclosed.

• · v4.4.4 release: Introduced a new benchmarking system for real-world coding tasks (enhances fairness and accuracy in agent evaluation).
• · King-of-the-Hill framework update: Optimized the matching algorithm for agent battles (reduces latency and improves scalability for larger miner pools).
• · Agent scoring model refinement: Improved metrics for evaluating code quality and correctness (ensures more reliable champion selection).
• · Open-source integration: Added support for third-party agent submission tools (lowers the barrier to entry for new miners and developers).

SN66 Ninja currently has a market cap of $16.3 million, placing it among mid-tier Bittensor subnets. Its 24-hour and 7-day price changes of -6.6% and -17.7%, respectively, indicate a bearish trend in the short term. The subnet's health score of 71/100 suggests moderate operational and technical stability, but the market seems to be pricing in uncertainty or slower adoption. Compared to similar subnets like SN17 (Deeper Learning) or SN33 (Vision), Ninja's focus on real-world coding tasks gives it a unique value proposition, but it remains to be seen if this differentiates it effectively in the broader market.

• ↑ Release of a major version with enhanced benchmarking and agent evaluation systems.
• ↑ Integration with popular open-source code repositories for more diverse and challenging tasks.
• ↑ Partnership with a major AI research lab or educational institution to validate and promote the subnet's framework.
• ↑ A surge in miner participation due to improved reward distribution or lower entry barriers.
• ↑ Adoption by enterprise developers or startups using the subnet for AI-powered code generation and testing.

• ↓ High concentration of development activity in a small team with no public hiring signals, increasing execution risk.
• ↓ The competitive nature of the 'king-of-the-hill' model may lead to a winner-takes-all scenario, reducing long-term miner diversity.
• ↓ The subnet's focus on coding agents may not attract a broad enough audience to sustain growth and adoption.
• ↓ Potential for centralization if a single agent or group dominates the leaderboard for extended periods.
• ↓ Lack of clear governance or community-driven decision-making could lead to disputes or reduced trust.

TrajectoryRL (SN11) is a Bittensor subnet focused on optimizing AI agent policies for real-world tasks, aiming to make autonomous agents more cost-effective, faster, and reliable. Miners on this subnet compete to develop the best reinforcement learning (RL) strategies that improve agent performance across diverse environments. The subnet addresses the challenge of deploying RL agents in practical settings where adaptability and efficiency are crucial. It leverages Bittensor’s decentralized framework to reward miners who contribute high-quality optimization solutions.

In the last 14 days, the TrajectoryRL team has been active, with 18 commits made over two days by two contributors. This indicates a concentrated effort on foundational improvements, particularly in optimizing RL agent performance for real-world deployment. The team appears to be prioritizing public development, with no clear signs of private work. Commit frequency and authorship suggest a focused development team, though the pace is moderate. No recent hiring announcements or major release cadence has been observed, but the current development velocity is stable.

• · Optimization Framework Improvements: The team has been refining the foundational RL optimization framework (18 commits in 2 days), which is critical for ensuring agents are reliable and efficient in real-world applications.
• · Miner Competition Logic: Updates to miner competition logic aim to improve how miners are evaluated based on the quality of their RL strategies, enhancing the subnet’s overall performance and fairness.
• · Adaptability Enhancements: Recent commits focus on improving agent adaptability to diverse environments, a key factor in real-world deployment success.

TrajectoryRL has a market cap of $16.3M, with a 7-day price increase of 9.8% and a 24-hour decline of -1.6%. The subnet’s price action shows recent positive momentum, suggesting growing interest from the market. Its health score of 76/100 indicates a reasonably healthy subnet, though not among the top performers. Compared to other Bittensor subnets focused on AI optimization, TrajectoryRL’s niche in RL agent performance gives it a differentiated position. The price appears to be slightly ahead of fundamentals, but the 7-day trend is bullish.

• ↑ Release of a new optimization benchmark to evaluate miner performance.
• ↑ Integration with a major autonomous robotics project for real-world testing.
• ↑ Onboarding of a prominent reinforcement learning researcher to the team.
• ↑ Launch of a public leaderboard to incentivize miner competition and transparency.

• ↓ Execution risk if the team fails to deliver on promised optimization improvements.
• ↓ Concentration risk due to a small number of active contributors and miners.
• ↓ Mechanism weakness if the evaluation criteria for miner competition are not robust enough to prevent gaming.

SN7 Allways is a universal transaction layer designed to streamline and optimize decentralized transaction validation. It operates as a foundational infrastructure for processing and verifying contributions across distributed systems. The subnet aims to reduce redundancy in processing and enhance the accuracy of contribution scoring, making it a critical component for scalable decentralized applications. Its architecture supports a wide range of transaction types and is adaptable to various use cases in the Bittensor ecosystem.

The subnet has shown high development velocity in the last 14 days, with 18 commits made in a single day by 10 distinct authors. This suggests strong team coordination and active collaboration. The recent commits focused on optimizing transaction validation and scoring mechanisms, indicating a clear prioritization of performance improvements. The team appears to be working in a public-first manner, with most of the work visible in public repositories. There are no signs of recent hiring sprees, but the current team's productivity is robust, and the release cadence is consistent with a product-focused development approach.

• · Transaction validation optimization: The subnet reduced redundant processing by 30% through algorithmic improvements (critical for scalability and cost efficiency).
• · Scoring mechanism update: A new scoring model was implemented to improve accuracy in contribution validation (enhances trust and reliability in the network).
• · Modular architecture enhancements: Introduced modular components for transaction handling to allow for easier integration with external systems (supports broader adoption and use cases).

SN7 Allways has a market cap of $5.3M, placing it in the mid-tier of Bittensor subnets. It has outperformed the broader market in the last 24 hours with a 7.6% price increase and a 14.2% gain over seven days. The price action suggests growing investor confidence, though the fundamentals are still catching up to the momentum. Its health score of 65/100 indicates that while the subnet is functional, there is room for improvement in terms of robustness and adoption. It is comparable to other infrastructure-focused subnets but lags slightly in terms of total value locked and user base.

• ↑ Launch of a public testnet to attract external developers and validators.
• ↑ Integration with a major decentralized finance (DeFi) platform to enable transaction layer usage in real-world applications.
• ↑ Partnership with a blockchain analytics firm to provide transparency and performance metrics for the subnet.

• ↓ High concentration of commits from a small group of developers increases execution risk if key contributors leave.
• ↓ The subnet's modular design may introduce complexity in long-term maintenance and upgrades.
• ↓ Competition from established transaction layer solutions in the broader blockchain space could hinder adoption.