With the amount of new subnets being added it can be hard to get up to date information across all subnets, so data may be slightly out of date from time to time
With the amount of new subnets being added it can be hard to get up to date information across all subnets, so data may be slightly out of date from time to time
Bittensor Subnet 20 “Bounty Hunter” is a specialized decentralized AI competition platform built on the Bittensor network. It transforms the traditional Bittensor “mining” model into a bounty-driven challenge system that incentivizes global AI developers to solve hard AI tasks. In essence, Subnet 20 hosts open AI competitions (or “bounties”) on benchmark challenges, automatically evaluates submissions, and rewards the top-performing models with cryptocurrency prizes. This subnet was previously known as BitAgent (sometimes informally referred to as “BitQuant”), which focused on LLM tool-calling agents. Now, as Bounty Hunter, it has pivoted to a broader mission: crowdsourcing breakthroughs on AI benchmarks by leveraging Bittensor’s token incentives. Key points about what SN20 does include:
Hosts AI Challenges: It integrates prominent AI benchmarks (e.g. Berkeley’s Function-Calling Leaderboard, Princeton’s SWE-Bench for code, Yale’s Spider 2.0 text-to-SQL, etc.) as competitions. Models are evaluated on these tasks in a transparent leaderboard format.
Incentivizes Open Participation: Anyone can participate by submitting models or solutions – no need to run a Bittensor miner node or stake upfront. This opens the door to global AI talent beyond the Bittensor community. Submissions are required to be open-source (Apache 2.0), ensuring transparency and enabling others to build on the results.
Automated Evaluation & Rewards: The subnet’s validator nodes automatically validate and score each submission against the benchmark (similar to Kaggle-style evaluation). Bittensor’s on-chain token emissions fund the bounties, which grow over time until someone achieves a winning result. Top performers receive TAO (detailed as SN20 “alpha” tokens) payouts, aligning incentives with model accuracy and innovation.
Overall, Bounty Hunter turns Bittensor into a decentralized “AI Kaggle”: instead of continuous Q&A mining, it runs targeted contests to push state-of-the-art on difficult AI problems. Team Rizzo emphasizes that Bittensor’s incentive system provides an edge over traditional AI competition platforms by continuously funding challenges and objectively rewarding solutions via blockchain. This makes Subnet 20 a “global engine for solving the hardest AI problems” using decentralized AI collaboration.
Bittensor Subnet 20 “Bounty Hunter” is a specialized decentralized AI competition platform built on the Bittensor network. It transforms the traditional Bittensor “mining” model into a bounty-driven challenge system that incentivizes global AI developers to solve hard AI tasks. In essence, Subnet 20 hosts open AI competitions (or “bounties”) on benchmark challenges, automatically evaluates submissions, and rewards the top-performing models with cryptocurrency prizes. This subnet was previously known as BitAgent (sometimes informally referred to as “BitQuant”), which focused on LLM tool-calling agents. Now, as Bounty Hunter, it has pivoted to a broader mission: crowdsourcing breakthroughs on AI benchmarks by leveraging Bittensor’s token incentives. Key points about what SN20 does include:
Hosts AI Challenges: It integrates prominent AI benchmarks (e.g. Berkeley’s Function-Calling Leaderboard, Princeton’s SWE-Bench for code, Yale’s Spider 2.0 text-to-SQL, etc.) as competitions. Models are evaluated on these tasks in a transparent leaderboard format.
Incentivizes Open Participation: Anyone can participate by submitting models or solutions – no need to run a Bittensor miner node or stake upfront. This opens the door to global AI talent beyond the Bittensor community. Submissions are required to be open-source (Apache 2.0), ensuring transparency and enabling others to build on the results.
Automated Evaluation & Rewards: The subnet’s validator nodes automatically validate and score each submission against the benchmark (similar to Kaggle-style evaluation). Bittensor’s on-chain token emissions fund the bounties, which grow over time until someone achieves a winning result. Top performers receive TAO (detailed as SN20 “alpha” tokens) payouts, aligning incentives with model accuracy and innovation.
Overall, Bounty Hunter turns Bittensor into a decentralized “AI Kaggle”: instead of continuous Q&A mining, it runs targeted contests to push state-of-the-art on difficult AI problems. Team Rizzo emphasizes that Bittensor’s incentive system provides an edge over traditional AI competition platforms by continuously funding challenges and objectively rewarding solutions via blockchain. This makes Subnet 20 a “global engine for solving the hardest AI problems” using decentralized AI collaboration.
Bounty Hunter’s product is a combination of an on-chain subnet protocol and an off-chain web platform (AIBoards.io) enabling AI challenge competitions. Technically, the “build” has two core components:
The Bittensor Subnet (SN20): Under the hood, SN20 operates as a Bittensor subnet (network UID 20) running custom logic for competition-based mining. It retains the classic miner/validator framework but repurposes it for benchmark tasks. Validator nodes in SN20 serve as automated judges – they fetch tasks from integrated datasets (e.g. BFCL tasks, SWE-Bench problems), invoke miners or submitted models on those tasks, and evaluate the outputs against ground-truth or test cases. This is analogous to an on-chain “scoring oracle.” Miner nodes (when used) host AI models that tackle the tasks, but notably Bounty Hunter also accepts one-time submissions from external participants (these are evaluated similarly by validators). The subnet’s blockchain logic allocates daily token emissions into a reward pool, rather than steady miner rewards, which accumulates until disbursed to winners. All models are constrained to be relatively lightweight (≤8B parameters) and open-source, so they can run on commodity GPUs and be audited by the community.
The AIBoards Web Platform: The front-end of Bounty Hunter is the AIBoards.io website, which provides a user interface for the competitions. On AIBoards, challenge descriptions and rules are posted, and participants can upload their solutions (e.g. model files, code, or other required outputs). The site outlines the process in steps – “Create & Post” a bounty (for challenge creators), “Hunt & Submit” for solvers to upload their work, and then “Earn” as automated validation determines the winners. Coming soon: the platform will even allow third-party users or organizations to create their own bounties with custom requirements, deadlines, and multi-winner criteria. Once a submission is made, the backend (via SN20 validators) automatically runs the solution through the benchmark tests and updates the leaderboard. Network integration: SN20 can also call on other Bittensor subnets for specialized skills – for example, routing a coding task to Subnet 45 (Team Rizzo’s SWE-Rizzo for code generation) as part of a pipeline. This cross-subnet orchestration allows Bounty Hunter to tackle multifaceted problems by composing different AI services.
In summary, the product is a decentralized AI contest platform: AIBoards provides the user-facing portal for bounties, while the SN20 subnet provides the trustless execution and evaluation engine. This architecture ensures that competition results are reproducible and tamper-proof (thanks to on-chain validation), and rewards are distributed transparently. Notably, Team Rizzo’s previous SN20 products (such as GoGoAgent.ai and MSPTech.ai) demonstrated the underlying tech – e.g. using tool-calling agents for IT automation – but the current build refocuses these capabilities into the Bounty Hunter platform. The result is an end-to-end system where AI models compete to solve tasks, and the best solutions are automatically verified and rewarded on-chain.
Bounty Hunter’s product is a combination of an on-chain subnet protocol and an off-chain web platform (AIBoards.io) enabling AI challenge competitions. Technically, the “build” has two core components:
The Bittensor Subnet (SN20): Under the hood, SN20 operates as a Bittensor subnet (network UID 20) running custom logic for competition-based mining. It retains the classic miner/validator framework but repurposes it for benchmark tasks. Validator nodes in SN20 serve as automated judges – they fetch tasks from integrated datasets (e.g. BFCL tasks, SWE-Bench problems), invoke miners or submitted models on those tasks, and evaluate the outputs against ground-truth or test cases. This is analogous to an on-chain “scoring oracle.” Miner nodes (when used) host AI models that tackle the tasks, but notably Bounty Hunter also accepts one-time submissions from external participants (these are evaluated similarly by validators). The subnet’s blockchain logic allocates daily token emissions into a reward pool, rather than steady miner rewards, which accumulates until disbursed to winners. All models are constrained to be relatively lightweight (≤8B parameters) and open-source, so they can run on commodity GPUs and be audited by the community.
The AIBoards Web Platform: The front-end of Bounty Hunter is the AIBoards.io website, which provides a user interface for the competitions. On AIBoards, challenge descriptions and rules are posted, and participants can upload their solutions (e.g. model files, code, or other required outputs). The site outlines the process in steps – “Create & Post” a bounty (for challenge creators), “Hunt & Submit” for solvers to upload their work, and then “Earn” as automated validation determines the winners. Coming soon: the platform will even allow third-party users or organizations to create their own bounties with custom requirements, deadlines, and multi-winner criteria. Once a submission is made, the backend (via SN20 validators) automatically runs the solution through the benchmark tests and updates the leaderboard. Network integration: SN20 can also call on other Bittensor subnets for specialized skills – for example, routing a coding task to Subnet 45 (Team Rizzo’s SWE-Rizzo for code generation) as part of a pipeline. This cross-subnet orchestration allows Bounty Hunter to tackle multifaceted problems by composing different AI services.
In summary, the product is a decentralized AI contest platform: AIBoards provides the user-facing portal for bounties, while the SN20 subnet provides the trustless execution and evaluation engine. This architecture ensures that competition results are reproducible and tamper-proof (thanks to on-chain validation), and rewards are distributed transparently. Notably, Team Rizzo’s previous SN20 products (such as GoGoAgent.ai and MSPTech.ai) demonstrated the underlying tech – e.g. using tool-calling agents for IT automation – but the current build refocuses these capabilities into the Bounty Hunter platform. The result is an end-to-end system where AI models compete to solve tasks, and the best solutions are automatically verified and rewarded on-chain.
Rizzo Founders
roguetensor – Co-founder & CTO: Tech visionary blending AI, Computer Vision, and Robotics to bring groundbreaking products to life. A catalyst for innovation and execution, bridging bold ideas with real-world applications.
frankrizz07 – Co-founder & CEO: Serial entrepreneur dedicated to solving complex, real-world challenges for executives—specializing in automation, recruitment/retention, and operational efficiency.
Rizzo Validator Team
rysjol – Data & Software Engineer: Experienced in network and system administration. Enthusiastic about geography, history, music, and sports.
solros3 – Technical Creative & VFX Artist: Combines a background in software engineering with VFX artistry, specializing in particle/fluid simulations and creative tech.
gregbeard – Full-Stack Developer & Systems Engineer: A versatile computer science expert, bringing robust skills in full-stack web development and system design.
Rizzo Subnet Development Team
slaive – Full-Stack Developer & Technologist: Fearless innovator with a knack for engineering, AI models, and the occasional high-risk experiment.
vectorforge – Subnet Developer, SN20 Specialist: Bringing strong experience in generative AI, data science, and backend systems to the forefront of SN20 development.
thunderheavyindustry – Application Developer & Polymath: Rooted in mathematics and philosophy, with talents in music and cuisine. Focused on integrating subnet outputs (SN20/SN45) into broader ecosystems.
brokespace – Cybersecurity & AI/ML Specialist: Focused on the intersection of cybersecurity, AI, and machine learning, especially in the context of large language models.
canti_dev – Generative AI Engineer: Veteran developer building end-to-end applications with emphasis on rapid prototyping and market-oriented solutions.
Rizzo Communications & Support Team
msptech.aiops – Operations & Strategy Leader: Bringing deep experience in team leadership and business growth across IT, logistics, manufacturing, and retail industries.
deakins02 – Digital Creative & Videographer: Award-winning visual storyteller with expertise in cinematography, animation, and digital editing.
taospark – Communications Strategist & AI Enthusiast: Computer scientist driven by a passion for automation, AI, and uniting teams to turn bold visions into tangible outcomes.
Rizzo Founders
roguetensor – Co-founder & CTO: Tech visionary blending AI, Computer Vision, and Robotics to bring groundbreaking products to life. A catalyst for innovation and execution, bridging bold ideas with real-world applications.
frankrizz07 – Co-founder & CEO: Serial entrepreneur dedicated to solving complex, real-world challenges for executives—specializing in automation, recruitment/retention, and operational efficiency.
Rizzo Validator Team
rysjol – Data & Software Engineer: Experienced in network and system administration. Enthusiastic about geography, history, music, and sports.
solros3 – Technical Creative & VFX Artist: Combines a background in software engineering with VFX artistry, specializing in particle/fluid simulations and creative tech.
gregbeard – Full-Stack Developer & Systems Engineer: A versatile computer science expert, bringing robust skills in full-stack web development and system design.
Rizzo Subnet Development Team
slaive – Full-Stack Developer & Technologist: Fearless innovator with a knack for engineering, AI models, and the occasional high-risk experiment.
vectorforge – Subnet Developer, SN20 Specialist: Bringing strong experience in generative AI, data science, and backend systems to the forefront of SN20 development.
thunderheavyindustry – Application Developer & Polymath: Rooted in mathematics and philosophy, with talents in music and cuisine. Focused on integrating subnet outputs (SN20/SN45) into broader ecosystems.
brokespace – Cybersecurity & AI/ML Specialist: Focused on the intersection of cybersecurity, AI, and machine learning, especially in the context of large language models.
canti_dev – Generative AI Engineer: Veteran developer building end-to-end applications with emphasis on rapid prototyping and market-oriented solutions.
Rizzo Communications & Support Team
msptech.aiops – Operations & Strategy Leader: Bringing deep experience in team leadership and business growth across IT, logistics, manufacturing, and retail industries.
deakins02 – Digital Creative & Videographer: Award-winning visual storyteller with expertise in cinematography, animation, and digital editing.
taospark – Communications Strategist & AI Enthusiast: Computer scientist driven by a passion for automation, AI, and uniting teams to turn bold visions into tangible outcomes.
Bounty Hunter’s roadmap is centered on expanding the range and impact of its AI competitions, while solidifying the platform for broader use. Key milestones and upcoming plans include:
Dominating Key Benchmarks: The team is targeting top rankings on several prestigious AI leaderboards through Bounty Hunter competitions. Recently, their 8-billion-parameter agent model reached #1 in its class on Berkeley’s Function-Calling Leaderboard (BFCL). Next, they are aiming for #1 overall on BFCL v4, which includes more difficult memory and web-interaction tasks (competing even against larger proprietary models). Similarly, for Princeton’s SWE-Bench (software engineering benchmark), Bounty Hunter will incorporate coding challenges (in collaboration with Subnet 45) to climb toward the top overall position.
New Bounty Integrations: The subnet will continuously add cutting-edge challenges as new “bounties.” For example, Team Rizzo announced plans to integrate Yale’s Spider 2.0 (an advanced text-to-SQL benchmark) as an upcoming competition as soon as that dataset is available. By adding Spider 2.0, Bounty Hunter will broaden into NLP and database-query tasks, pushing models to handle complex text-to-database reasoning. Expect further integrations of AI benchmarks across domains (language understanding, reasoning, multimodal tasks, etc.) as the platform grows.
Launch and Enhancement of AIBoards Platform: In October 2025, the Bittensor Bounty Hunter frontend (AIBoards.io) was launched to the public. Over the coming weeks, Team Rizzo plans to refine this platform, adding features like user accounts (wallet-linked), detailed leaderboards, and possibly a reputation system for repeat participants. A major upcoming feature is enabling external AI stakeholders to post their own bounties – the “Creators post bounties” functionality is listed as “Coming Soon” on the site. This will allow companies, researchers, or community members to sponsor challenges on Bounty Hunter, accelerating the variety of tasks and real-world problems tackled.
Growing the Reward Pool & Incentives: Thanks to dTAO emissions and staking support, Subnet 20’s bounty reward pool is substantial and will continue to grow. As of mid-2025 the pool had reached the equivalent of ~$130k (over 85,000 SN20 tokens) and has likely increased since. The roadmap includes dynamic bounty scaling – if a challenge remains unsolved or no one beats the current best score, the prize pot for that bounty will keep increasing over time. This mechanism is designed to continuously incentivize breakthroughs and attract top talent to take on the hardest problems.
Long-Term Vision – “Global AI Bounty Network”: Team Rizzo envisions Bounty Hunter as just the start. In 2025 they described this pivot as creating a global engine for solving AI problems, surpassing Kaggle-like contests by leveraging crypto incentives. By 2026 and beyond, the goal is to achieve global reach – engaging thousands of AI experts around the world and integrating Bounty Hunter into academic and enterprise AI workflows. They also foresee deeper cross-subnet interoperability in Bittensor: for example, Bounty Hunter might coordinate with future subnets (for vision, audio, etc.) to launch multi-modal challenges, all while using the same decentralized reward system. This aligns with Team Rizzo’s broader mission of “democratizing AI through decentralized networks and community-driven innovation,” ensuring the hardest AI challenges can be openly tackled by the community.
In summary, the roadmap for Subnet 20 – Bounty Hunter focuses on scaling up competitions in number and difficulty, improving the platform for wider participation, and continuously pushing the frontier of AI performance. Each success on a leaderboard or new bounty added not only validates the subnet’s approach but also strengthens Bittensor’s position as a hub for collaborative AI development. Team Rizzo’s ambitious plan is to turn SN20 into a self-sustaining AI innovation engine – where progress in AI is accelerated by decentralized incentives, and where anyone, anywhere can contribute to and benefit from the cutting edge of machine learning.
Bounty Hunter’s roadmap is centered on expanding the range and impact of its AI competitions, while solidifying the platform for broader use. Key milestones and upcoming plans include:
Dominating Key Benchmarks: The team is targeting top rankings on several prestigious AI leaderboards through Bounty Hunter competitions. Recently, their 8-billion-parameter agent model reached #1 in its class on Berkeley’s Function-Calling Leaderboard (BFCL). Next, they are aiming for #1 overall on BFCL v4, which includes more difficult memory and web-interaction tasks (competing even against larger proprietary models). Similarly, for Princeton’s SWE-Bench (software engineering benchmark), Bounty Hunter will incorporate coding challenges (in collaboration with Subnet 45) to climb toward the top overall position.
New Bounty Integrations: The subnet will continuously add cutting-edge challenges as new “bounties.” For example, Team Rizzo announced plans to integrate Yale’s Spider 2.0 (an advanced text-to-SQL benchmark) as an upcoming competition as soon as that dataset is available. By adding Spider 2.0, Bounty Hunter will broaden into NLP and database-query tasks, pushing models to handle complex text-to-database reasoning. Expect further integrations of AI benchmarks across domains (language understanding, reasoning, multimodal tasks, etc.) as the platform grows.
Launch and Enhancement of AIBoards Platform: In October 2025, the Bittensor Bounty Hunter frontend (AIBoards.io) was launched to the public. Over the coming weeks, Team Rizzo plans to refine this platform, adding features like user accounts (wallet-linked), detailed leaderboards, and possibly a reputation system for repeat participants. A major upcoming feature is enabling external AI stakeholders to post their own bounties – the “Creators post bounties” functionality is listed as “Coming Soon” on the site. This will allow companies, researchers, or community members to sponsor challenges on Bounty Hunter, accelerating the variety of tasks and real-world problems tackled.
Growing the Reward Pool & Incentives: Thanks to dTAO emissions and staking support, Subnet 20’s bounty reward pool is substantial and will continue to grow. As of mid-2025 the pool had reached the equivalent of ~$130k (over 85,000 SN20 tokens) and has likely increased since. The roadmap includes dynamic bounty scaling – if a challenge remains unsolved or no one beats the current best score, the prize pot for that bounty will keep increasing over time. This mechanism is designed to continuously incentivize breakthroughs and attract top talent to take on the hardest problems.
Long-Term Vision – “Global AI Bounty Network”: Team Rizzo envisions Bounty Hunter as just the start. In 2025 they described this pivot as creating a global engine for solving AI problems, surpassing Kaggle-like contests by leveraging crypto incentives. By 2026 and beyond, the goal is to achieve global reach – engaging thousands of AI experts around the world and integrating Bounty Hunter into academic and enterprise AI workflows. They also foresee deeper cross-subnet interoperability in Bittensor: for example, Bounty Hunter might coordinate with future subnets (for vision, audio, etc.) to launch multi-modal challenges, all while using the same decentralized reward system. This aligns with Team Rizzo’s broader mission of “democratizing AI through decentralized networks and community-driven innovation,” ensuring the hardest AI challenges can be openly tackled by the community.
In summary, the roadmap for Subnet 20 – Bounty Hunter focuses on scaling up competitions in number and difficulty, improving the platform for wider participation, and continuously pushing the frontier of AI performance. Each success on a leaderboard or new bounty added not only validates the subnet’s approach but also strengthens Bittensor’s position as a hub for collaborative AI development. Team Rizzo’s ambitious plan is to turn SN20 into a self-sustaining AI innovation engine – where progress in AI is accelerated by decentralized incentives, and where anyone, anywhere can contribute to and benefit from the cutting edge of machine learning.
Oh yeah, I forgot to tell you.
SN45 went LIVE on Main Net today 🥳🥳🥳
Bittensor Subnet 45 is the engine powering Talisman’s agentic-AI features.
Together with Rizzo, we've designed a performance-driven SN45 buyback mechanism that aligns subnet incentives with Talisman’s growth.
A flywheel where users win, miners win, and the network compounds.
Join us next Tuesday for SN45 Agentic Alliance, an in-depth AMA covering our strategic partnership w/ @TeamRizzoAI
We’ll be going under the hood of Bittensor Subnet 45, the engine behind Talisman’s agentic AI infrastructure.
If you're attending DevConnect next week, join me as I present;
1. Why Bittensor beats traditional business
2. The Future of Real Estate as @resilabsai scales
Thank you @ThisIsDNA and @TeamRizzoAI for the invite.
@CryptoZPunisher @TeamRizzoAI Funny how people stare at squiggly lines hoping to see the future when the real treasure is in the code and community behind the token. Charts show where it’s been; fundamentals whisper where it’s going. SN45 feels more like detective work than fortune-telling, doesn’t it?
Big things coming to Talisman!
Congrats team 👏
Scheduled a Revenue Search episode w @SiamKidd on Nov 26th.
We're going to dive into Talisman AI - SN45!!
I might even drop a 💣
#Bittensor
$TAO
