principia-ai/PhysGym
A benchmark suite for evaluating LLM-based interactive scientific reasoning.
This project helps AI researchers and scientists rigorously test how well large language models (LLMs) can discover physics laws. It takes an LLM agent and a physics problem, systematically controlling what prior information the agent receives (from full context to anonymous variables). The output shows how successfully the LLM deduces or experiments to find the correct physics equation, revealing whether it's memorizing or truly reasoning.
Use this if you are an AI researcher or cognitive scientist developing or evaluating LLM agents for scientific discovery and need to understand how different levels of prior knowledge impact their reasoning abilities.
Not ideal if you are looking for an off-the-shelf tool to solve practical physics problems or to apply LLMs in an industrial setting without deep research into their discovery capabilities.
Stars
92
Forks
12
Language
Python
License
MIT
Category
Last pushed
Jan 06, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/llm-tools/principia-ai/PhysGym"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Featured in
Higher-rated alternatives
sierra-research/tau2-bench
τ²-Bench: Evaluating Conversational Agents in a Dual-Control Environment
xlang-ai/OSWorld
[NeurIPS 2024] OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments
bigcode-project/bigcodebench
[ICLR'25] BigCodeBench: Benchmarking Code Generation Towards AGI
THUDM/AgentBench
A Comprehensive Benchmark to Evaluate LLMs as Agents (ICLR'24)
scicode-bench/SciCode
A benchmark that challenges language models to code solutions for scientific problems