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of new subnets being added it can be hard to get up to date information across all subnets, so data may be slightly out of date from time to time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of new subnets being added it can be hard to get up to date information across all subnets, so data may be slightly out of date from time to time
The Brain subnet (SN90) is a specialized Bittensor subnet for truth verification in prediction markets. According to its creators, Brain “is designed to create a decentralized prediction market verification system”. In practice, Brain miners are given factual statements or claims and must determine whether each is true or false (often with a confidence score). Validators then check the miners’ work for accuracy and sound methodology. The goal is to generate a reliable “truth oracle” – a digital commodity of verified statements – within the Bittensor ecosystem. More accurate miners (and validators) earn a larger share of TAO token emissions, providing incentive alignment. In Bittensor’s network of subnets (each focusing on a distinct AI task), Brain occupies the niche of fact-checking/prediction validation.
The Brain subnet (SN90) is a specialized Bittensor subnet for truth verification in prediction markets. According to its creators, Brain “is designed to create a decentralized prediction market verification system”. In practice, Brain miners are given factual statements or claims and must determine whether each is true or false (often with a confidence score). Validators then check the miners’ work for accuracy and sound methodology. The goal is to generate a reliable “truth oracle” – a digital commodity of verified statements – within the Bittensor ecosystem. More accurate miners (and validators) earn a larger share of TAO token emissions, providing incentive alignment. In Bittensor’s network of subnets (each focusing on a distinct AI task), Brain occupies the niche of fact-checking/prediction validation.
Brain follows Bittensor’s standard miner–validator architecture. In each “round,” a validator issues a batch of statements to the network. Participating miners then attempt to verify each statement using any open-source method they choose (e.g. web searches, database lookups, logical reasoning, external AI models). Each miner returns a structured answer: a true/false determination with a confidence score (0.0–1.0), along with supporting evidence or explanation. For example, a miner might respond with “False – 95% confidence” plus a URL or text excerpt as evidence. Validators then independently evaluate all submitted answers. They score each miner on criteria such as correctness (compared to ground truth when known), evidence quality, explanation clarity, and methodological soundness. After reviewing, validators distribute TAO rewards to miners proportional to their accuracy and thoroughness. In addition, validators implement a cross-validation step: they may send a miner’s answer back to other miners to double-check the work. These secondary miners can flag errors or provide alternative answers. This cross-check process helps prevent cheating and reinforces integrity in the network.
Miners and validators post their performance scores on-chain. Each validator produces a score for every miner over recent epochs; the resulting score matrix is fed into Bittensor’s Yuma consensus. The on-chain consensus algorithm then allocates TAO emissions to miners, validators, and the subnet owner based on this performance data. In effect, high-performing miners and diligent validators receive more TAO rewards. The Brain subnet’s incentive mechanism is explicitly designed so that “the most accurate and reliable verification miners receive a greater share of TAO emissions”. In summary, Brain’s workflow is: (1) Validators issue statements; (2) Miners verify and return answers (true/false, confidence, evidence); (3) Validators score the answers and may have others double-check results; (4) Validators distribute rewards based on those scores. All of this runs as a Bittensor subnet protocol off-chain (in Python) but integrated on-chain via Bittensor’s staking and consensus mechanisms.
Products and Tools: Brain does not (yet) offer a public web app or API; its deliverable is primarily the verification service itself. The core software is an open-source Python package/repository (on GitHub) that implements the subnet’s incentive logic and neuron types. It provides command-line tools for participants. For example, users can run a miner with the CLI command brain-miner –netuid 90 –wallet.name <name> –wallet.hotkey <key> (similarly brain-validator) to join Brain on the Bittensor network. These tools are analogous to other Bittensor subnet CLI utilities, customized for Brain’s task. Brain participants must run the standard Bittensor node software along with these Python modules, then the subnet’s protocol handles queries internally.
Brain’s alpha token is tradeable on Bittensor markets (as “dTAO – Brain”) once the subnet is active. Early on, liquidity is extremely low, but the token is listed on interfaces like TAO.app or Backprop’s terminal. Community-driven projects have already planned integrations: for instance, the AI prediction-market platform DegenPredict announced a “Brain Subnet Rewards Program” in which holders of the Brain token (α) will receive DGEN token airdrops and revenue share, starting mid-May 2025. This indicates growing ecosystem support: developers can build on Brain’s outputs (verified statements) or simply trade/stake its token. Beyond the CLI and blockchain integration, there is no separate user-facing “Brain app.” The repository and documentation serve as the primary technical resource (see below).
Technical Architecture: Brain is built as a standard Bittensor subnet. Its codebase uses the Bittensor Python SDK and is structured into two main parts: the protocol/reward logic and the neuron implementations. In the GitHub repo, the brain/ directory defines the core protocol (with files like protocol.py, forward.py, reward.py) and the neurons/ directory contains miner.py and validator.py. This follows the template used by Bittensor subnets: protocol.py defines how miners and validators communicate, forward.py likely handles the forward-pass of statements through the network, and reward.py implements the accuracy-based reward mechanism. The miner (BrainMiner) and validator (BrainValidator) classes implement the specific task logic (including the perform_verification() method miners override). Custom miners can inherit from BrainMiner to plug in their own fact-checking algorithms or models.
In practice, Brain nodes run on commodity hardware like any Bittensor node. There is no specialized hardware requirement listed; miners and validators could use generic CPUs/GPUs. The underlying chain is the Subtensor (Bittensor) blockchain. Brain uses a unique netuid (90) for identification in the network. During operation, validators publish weight updates and performance data on the Subtensor chain. Once the 7-day launch-lock is over, Yuma consensus on the Bittensor chain will issue actual TAO transfers according to those weights.
No specific ML model is dictated by the protocol – miners might use search engines or language models to fact-check statements, but the Brain code itself does not bundle an AI model. It relies on “open-source verification methods,” meaning miners can plug in whatever tools they prefer (e.g. GPT-based QA, specialized databases, etc.). All communication between nodes (miner queries, validator tasks) uses Bittensor’s off-chain networking. The deployment stack is essentially a Python environment with the Bittensor framework installed. Brain’s repository does not publish releases yet (it’s source-only), and it is licensed under MIT. In summary, Brain sits on the Bittensor decentralized infrastructure, implements its custom incentive protocol in Python, and interacts with the Bittensor blockchain for staking, scoring, and token distribution.
Brain follows Bittensor’s standard miner–validator architecture. In each “round,” a validator issues a batch of statements to the network. Participating miners then attempt to verify each statement using any open-source method they choose (e.g. web searches, database lookups, logical reasoning, external AI models). Each miner returns a structured answer: a true/false determination with a confidence score (0.0–1.0), along with supporting evidence or explanation. For example, a miner might respond with “False – 95% confidence” plus a URL or text excerpt as evidence. Validators then independently evaluate all submitted answers. They score each miner on criteria such as correctness (compared to ground truth when known), evidence quality, explanation clarity, and methodological soundness. After reviewing, validators distribute TAO rewards to miners proportional to their accuracy and thoroughness. In addition, validators implement a cross-validation step: they may send a miner’s answer back to other miners to double-check the work. These secondary miners can flag errors or provide alternative answers. This cross-check process helps prevent cheating and reinforces integrity in the network.
Miners and validators post their performance scores on-chain. Each validator produces a score for every miner over recent epochs; the resulting score matrix is fed into Bittensor’s Yuma consensus. The on-chain consensus algorithm then allocates TAO emissions to miners, validators, and the subnet owner based on this performance data. In effect, high-performing miners and diligent validators receive more TAO rewards. The Brain subnet’s incentive mechanism is explicitly designed so that “the most accurate and reliable verification miners receive a greater share of TAO emissions”. In summary, Brain’s workflow is: (1) Validators issue statements; (2) Miners verify and return answers (true/false, confidence, evidence); (3) Validators score the answers and may have others double-check results; (4) Validators distribute rewards based on those scores. All of this runs as a Bittensor subnet protocol off-chain (in Python) but integrated on-chain via Bittensor’s staking and consensus mechanisms.
Products and Tools: Brain does not (yet) offer a public web app or API; its deliverable is primarily the verification service itself. The core software is an open-source Python package/repository (on GitHub) that implements the subnet’s incentive logic and neuron types. It provides command-line tools for participants. For example, users can run a miner with the CLI command brain-miner –netuid 90 –wallet.name <name> –wallet.hotkey <key> (similarly brain-validator) to join Brain on the Bittensor network. These tools are analogous to other Bittensor subnet CLI utilities, customized for Brain’s task. Brain participants must run the standard Bittensor node software along with these Python modules, then the subnet’s protocol handles queries internally.
Brain’s alpha token is tradeable on Bittensor markets (as “dTAO – Brain”) once the subnet is active. Early on, liquidity is extremely low, but the token is listed on interfaces like TAO.app or Backprop’s terminal. Community-driven projects have already planned integrations: for instance, the AI prediction-market platform DegenPredict announced a “Brain Subnet Rewards Program” in which holders of the Brain token (α) will receive DGEN token airdrops and revenue share, starting mid-May 2025. This indicates growing ecosystem support: developers can build on Brain’s outputs (verified statements) or simply trade/stake its token. Beyond the CLI and blockchain integration, there is no separate user-facing “Brain app.” The repository and documentation serve as the primary technical resource (see below).
Technical Architecture: Brain is built as a standard Bittensor subnet. Its codebase uses the Bittensor Python SDK and is structured into two main parts: the protocol/reward logic and the neuron implementations. In the GitHub repo, the brain/ directory defines the core protocol (with files like protocol.py, forward.py, reward.py) and the neurons/ directory contains miner.py and validator.py. This follows the template used by Bittensor subnets: protocol.py defines how miners and validators communicate, forward.py likely handles the forward-pass of statements through the network, and reward.py implements the accuracy-based reward mechanism. The miner (BrainMiner) and validator (BrainValidator) classes implement the specific task logic (including the perform_verification() method miners override). Custom miners can inherit from BrainMiner to plug in their own fact-checking algorithms or models.
In practice, Brain nodes run on commodity hardware like any Bittensor node. There is no specialized hardware requirement listed; miners and validators could use generic CPUs/GPUs. The underlying chain is the Subtensor (Bittensor) blockchain. Brain uses a unique netuid (90) for identification in the network. During operation, validators publish weight updates and performance data on the Subtensor chain. Once the 7-day launch-lock is over, Yuma consensus on the Bittensor chain will issue actual TAO transfers according to those weights.
No specific ML model is dictated by the protocol – miners might use search engines or language models to fact-check statements, but the Brain code itself does not bundle an AI model. It relies on “open-source verification methods,” meaning miners can plug in whatever tools they prefer (e.g. GPT-based QA, specialized databases, etc.). All communication between nodes (miner queries, validator tasks) uses Bittensor’s off-chain networking. The deployment stack is essentially a Python environment with the Bittensor framework installed. Brain’s repository does not publish releases yet (it’s source-only), and it is licensed under MIT. In summary, Brain sits on the Bittensor decentralized infrastructure, implements its custom incentive protocol in Python, and interacts with the Bittensor blockchain for staking, scoring, and token distribution.
The Brain project appears to be driven by a small team, primarily Fred Krueger (GitHub: frkrueger). Fred Krueger is a well-known pseudonymous crypto developer and author (co-author of The Big Bitcoin Book) who recently announced he “created Subnet 90 on Bittensor: Brain”. He committed the initial code on Apr 5, 2025 (as shown by the GitHub history). No other committers are listed in the repo, so Fred is likely the lead and sole initial implementer. On social media, several community members have engaged with Brain. For example, the Twitter users Punisher ττ and TAOisTheKey have publicly commented on Brain’s launch and token, and Fred himself tweeted about it. However, these appear to be community/investor contributors rather than additional developers. (No public organization or additional GitHub contributors are evident at this time.) Fred Krueger also has an active presence on Twitter/X (@dotkrueger) and a personal site where he listed Brain as a recent creation. The repository itself has been starred and watched by a handful of users, indicating initial community interest. Overall, Brain’s development team is largely pseudonymous; Fred Krueger is the only identified author, while others remain informal participants.
The Brain project appears to be driven by a small team, primarily Fred Krueger (GitHub: frkrueger). Fred Krueger is a well-known pseudonymous crypto developer and author (co-author of The Big Bitcoin Book) who recently announced he “created Subnet 90 on Bittensor: Brain”. He committed the initial code on Apr 5, 2025 (as shown by the GitHub history). No other committers are listed in the repo, so Fred is likely the lead and sole initial implementer. On social media, several community members have engaged with Brain. For example, the Twitter users Punisher ττ and TAOisTheKey have publicly commented on Brain’s launch and token, and Fred himself tweeted about it. However, these appear to be community/investor contributors rather than additional developers. (No public organization or additional GitHub contributors are evident at this time.) Fred Krueger also has an active presence on Twitter/X (@dotkrueger) and a personal site where he listed Brain as a recent creation. The repository itself has been starred and watched by a handful of users, indicating initial community interest. Overall, Brain’s development team is largely pseudonymous; Fred Krueger is the only identified author, while others remain informal participants.
Brain’s public development is at a very early stage. The GitHub “Initial commit” was made on April 5, 2025, and around that time community members noted its deployment. After launch, Bittensor subnets typically observe a mandatory 7-day lock before emission; Brain’s token (dTAO – Brain) shows zero emissions initially. Community commentary suggests Brain became available (token listed) in mid-April 2025 (it was “7 hours old” in one tweet) and is entering its active phase. As of mid-May 2025, Brain is presumably concluding its start-up phase. In fact, the DegenPredict platform’s rewards program notes that Brain holders will begin receiving airdrops and revenue shares starting in mid-May 2025, implying the subnet will be fully active by then.
Beyond this, no formal roadmap or milestones have been published on Brain’s GitHub or official channels. The main known next steps are: (a) commence TAO emissions once the launch lock expires, (b) attract miners and validators to participate in verifying statements, and (c) integrate Brain’s outputs with the wider ecosystem (as exemplified by the DegenPredict announcement). Any further technical enhancements (e.g. refining the verification algorithm, adding new features) have not been publicly detailed. In summary, Brain’s completed milestones include code development and subnet registration (early April 2025); current status is “launched, pending activation”; and upcoming goals are the start of official mining rewards and community adoption as a decentralized fact-checking market.
Brain’s public development is at a very early stage. The GitHub “Initial commit” was made on April 5, 2025, and around that time community members noted its deployment. After launch, Bittensor subnets typically observe a mandatory 7-day lock before emission; Brain’s token (dTAO – Brain) shows zero emissions initially. Community commentary suggests Brain became available (token listed) in mid-April 2025 (it was “7 hours old” in one tweet) and is entering its active phase. As of mid-May 2025, Brain is presumably concluding its start-up phase. In fact, the DegenPredict platform’s rewards program notes that Brain holders will begin receiving airdrops and revenue shares starting in mid-May 2025, implying the subnet will be fully active by then.
Beyond this, no formal roadmap or milestones have been published on Brain’s GitHub or official channels. The main known next steps are: (a) commence TAO emissions once the launch lock expires, (b) attract miners and validators to participate in verifying statements, and (c) integrate Brain’s outputs with the wider ecosystem (as exemplified by the DegenPredict announcement). Any further technical enhancements (e.g. refining the verification algorithm, adding new features) have not been publicly detailed. In summary, Brain’s completed milestones include code development and subnet registration (early April 2025); current status is “launched, pending activation”; and upcoming goals are the start of official mining rewards and community adoption as a decentralized fact-checking market.
Keep ahead of the Bittensor exponential development curve…
Subnet Alpha is an informational platform for Bittensor Subnets.
This site is not affiliated with the Opentensor Foundation or TaoStats.
The content provided on this website is for informational purposes only. We make no guarantees regarding the accuracy or currency of the information at any given time.
Subnet Alpha is created and maintained by The Realistic Trader. If you have any suggestions or encounter any issues, please contact us at [email protected].
Copyright 2024