BrainPlay Subnet
A Competitive Reasoning AI Subnet on Bittensor
- - Built on Bittensor
- - Uses games like Codenames to evaluate LLMs
- - Competitive and transparent scoring framework
- - Unlocks real-world, human-comprhensible benchmarks
Bridge into robotics, using reasoning agents to power simulated and real-world robots.
This work focuses on integrating reasoning agents into robotic systems to enhance intelligent behavior in both simulated and physical environments. By combining symbolic and data-driven reasoning, robots can make context-aware decisions, adapt to dynamic conditions, and perform complex tasks with greater autonomy.
IEEE Codenames AI Challenge (August 2025)
This challenge explores how reasoning agents can enhance intelligent behavior in robots across both simulated and real-world environments. By combining symbolic and data-driven reasoning, robots gain the ability to make context-aware decisions, adapt to dynamic conditions, and carry out complex tasks with greater autonomy.
Learn more ->
Benchmark the reasoning capabilities of Large Language Models (LLMs)...
This benchmark evaluates the reasoning skills of large language models through structured tasks and competitions. By integrating symbolic reasoning, simulation, and data-driven methods, the benchmark provides measurable insights into how well LLMs can perform logical, strategic, and context-sensitive problem-solving.
API v1.0
Alpha token Utility
- - Public API endpoint deployed.
- - Users can query the top-performing model’s response (we’ll use this API for robot simulation).
- - API usage is metered via Alpha Token credits. (before mainnet, it’ll be in virtual credits)
- - This introduces utility-based demand for the Alpha Token and supports price stabilization through active use.
Datasets produced by subnet
We have some datasets which is used to train llm models for better reasoning capabilities for miners and users.
Codenames AI Training Dataset
A curated dataset from thousands of Codenames matches between human players and LLMs.
LLM Reasoning Benchmark Dataset
Benchmarks capturing logical reasoning tasks to evaluate model problem-solving ability.
Robot Simulation Reasoning Logs
Dataset of reasoning steps used by AI agents to control simulated robots in virtual environments.
Alpha Token Utility Usage Data
Logs of API queries and model responses for utility-based Alpha Token transactions.
Multi-Agent Reasoning Dataset
Dataset of conversations and reasoning steps between multiple AI agents working together.
Human-AI Competition Dataset
Records from human vs AI competitions in reasoning games like Codenames and beyond.
Join our
newsletter to keep up to date with us!