TheoPhysEval: Assessing AI agents at autonomous theoretical physics problem solving
hep-th, cond-mat.stat-mech, cond-mat.str-el, cs.AI, gr-qc
Submitted: 2026-10-09
Updated: 2026-10-09
Code: https://github.com/anthropics/formal-math
Terminology
Sources
- Information Transfer with a Gravitating Bath
- Textured phase diagrams of featureless insulators
- Mathematical exploration and discovery at scale
- Resolution of Erd\H{o}s Problem #728: a writeup of Aristotle's Lean proof
- Remarks on the disproof of the unit distance conjecture
- A counterexample to a conjecture of Thakur on Carlitz-Wieferich primes
- A proof of an identity for the critical exponents of jamming
- Using Large Language Model to Solve and Explain Physics Word Problems Approaching Human Level
- PHYBench: Holistic Evaluation of Physical Perception and Reasoning in Large Language Models
- HiPhO: How Far Are (M)LLMs from Humans in the Latest High School Physics Olympiad Benchmark?
- Theoretical Physics Benchmark (TPBench) -- a Dataset and Study of AI Reasoning Capabilities in Theoretical Physics
- FrontierScience: Evaluating AI's Ability to Perform Expert-Level Scientific Tasks
- PRL-Bench: A Comprehensive Benchmark Evaluating LLMs' Capabilities in Frontier Physics Research
- QMBench: A Research Level Benchmark for Quantum Materials Research
- Probing the Critical Point (CritPt) of AI Reasoning: a Frontier Physics Research Benchmark
- First Proof
- First Proof Second Batch
- Aletheia tackles FirstProof autonomously
Related papers
- S-duality, boundary states, and higher-form symmetries on ALE spaces
- Quantum simulation of traversable-wormhole-inspired quantum teleportation in a chaotic binary sparse SYK model
- Perfect Particle Transmission through Duality Defects
- Entanglement Wedge Reconstruction Beyond the Large N Limit via the Twirled Petz Map
- Horizons and Soft Quantum Information
- Energy Transmission Across Holographic Conformal Interfaces in General Dimensions