# FrontierPhysics leaderboard

external benchmark

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelchoice.dev/leaderboards/benchmarks/external-frontier-physics

End-to-End Physics Research Benchmark for AI Agents. Every model the catalog carries a published FrontierPhysics value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | external |
| Measure | Agentic research in a sandbox: literature review, simulation or measurement, structured results and a written paper |
| Tasks | 56 research projects from 42 physicists across nine fields, three trials each (168 rollouts per setup) |
| Difficulty | Open-ended research projects that took their authors about 270 days on average |
| Published by | [FrontierPhysics: Evaluating Agents for End-to-End Physics Research (arXiv, November 4, 2026)](https://arxiv.org/abs/2611.06910) |

FrontierPhysics, from BenchFlow, gives an AI agent a research project written by a working physicist and checks whether it gets the physics right and writes it up well enough to publish. It has 56 tasks across nine fields of physics, and each one is run three times.

## FrontierPhysics ranking

28 models with a published FrontierPhysics value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.

| Rank | Model | Provider | Agentic research in a sandbox: literature review, simulation or measurement, structured results and a written paper | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Claude Fable 5.2](https://modelchoice.dev/models/claude-fable-5-2) | Anthropic | 33.3 | Observed 2026-11-21 · source benchlm:models |
| 2 | [Claude Opus 5.5](https://modelchoice.dev/models/claude-opus-5-5) | Anthropic | 27.4 | Observed 2026-11-21 · source benchlm:models |
| 2 | [Claude Sonnet 5.5](https://modelchoice.dev/models/claude-sonnet-5-5) | Anthropic | 27.4 | Observed 2026-11-21 · source benchlm:models |
| 4 | [GPT-6.2 Astra](https://modelchoice.dev/models/gpt-6-2-astra) | OpenAI | 26.8 | Observed 2026-11-21 · source benchlm:models |
| 5 | [Gemini 4 Neon](https://modelchoice.dev/models/gemini-4-neon) | Google | 23.2 | Observed 2026-11-21 · source benchlm:models |
| 5 | [GPT-6 Astra](https://modelchoice.dev/models/gpt-6-astra) | OpenAI | 23.2 | Observed 2026-11-21 · source benchlm:models |
| 7 | [GPT-6.2 Sol](https://modelchoice.dev/models/gpt-6-2-sol) | OpenAI | 21.4 | Observed 2026-11-21 · source benchlm:models |
| 8 | [GPT-6.1 Sol](https://modelchoice.dev/models/gpt-6-1-sol) | OpenAI | 19.6 | Observed 2026-11-21 · source benchlm:models |
| 9 | [Gemini 4 Carbon](https://modelchoice.dev/models/gemini-4-carbon) | Google | 19.0 | Observed 2026-11-21 · source benchlm:models |
| 10 | [GPT-6 Sol](https://modelchoice.dev/models/gpt-6-sol) | OpenAI | 14.3 | Observed 2026-11-21 · source benchlm:models |
| 10 | [GPT-6.1 Terra](https://modelchoice.dev/models/gpt-6-1-terra) | OpenAI | 14.3 | Observed 2026-11-21 · source benchlm:models |
| 12 | [Claude Fable 5.1](https://modelchoice.dev/models/claude-fable-5-1) | Anthropic | 11.3 | Observed 2026-11-21 · source benchlm:models |
| 13 | [Claude Opus 5](https://modelchoice.dev/models/claude-opus-5) | Anthropic | 9.5 | Observed 2026-11-21 · source benchlm:models |
| 14 | [Gemini 4 Argon](https://modelchoice.dev/models/gemini-4-argon) | Google | 8.3 | Observed 2026-11-21 · source benchlm:models |
| 15 | [GPT-5.6 Sol](https://modelchoice.dev/models/gpt-5-6-sol) | OpenAI | 7.1 | Observed 2026-11-21 · source benchlm:models |
| 16 | [Kimi K3.1 Pro](https://modelchoice.dev/models/kimi-3-1-pro) | Moonshot AI | 6.5 | Observed 2026-11-21 · source benchlm:models |
| 17 | [Claude Haiku 5.5](https://modelchoice.dev/models/claude-haiku-5-5) | Anthropic | 4.2 | Observed 2026-11-21 · source benchlm:models |
| 17 | [GLM-5.3](https://modelchoice.dev/models/glm-5-3) | Z.AI | 4.2 | Observed 2026-11-21 · source benchlm:models |
| 17 | [Grok 4.7](https://modelchoice.dev/models/grok-4-7) | xAI | 4.2 | Observed 2026-11-21 · source benchlm:models |
| 20 | [DeepSeek V4.1 Flash 1017](https://modelchoice.dev/models/deepseek-v4-1-flash) | DeepSeek | 3.6 | Observed 2026-11-21 · source benchlm:models |
| 20 | [GPT-5.6 Luna](https://modelchoice.dev/models/gpt-5-6-luna) | OpenAI | 3.6 | Observed 2026-11-21 · source benchlm:models |
| 20 | [GPT-5.6 Terra](https://modelchoice.dev/models/gpt-5-6-terra) | OpenAI | 3.6 | Observed 2026-11-21 · source benchlm:models |
| 23 | [Inkling 2 Small](https://modelchoice.dev/models/inkling-2-small) | Thinking Machines Lab | 3.0 | Observed 2026-11-21 · source benchlm:models |
| 24 | [Grok 4.6](https://modelchoice.dev/models/grok-4-6) | xAI | 2.4 | Observed 2026-11-21 · source benchlm:models |
| 24 | [Kimi K3](https://modelchoice.dev/models/kimi-3) | Moonshot AI | 2.4 | Observed 2026-11-21 · source benchlm:models |
| 26 | [MiMo-V2.6-Pro](https://modelchoice.dev/models/mimo-v2-6-pro) | Xiaomi | 1.2 | Observed 2026-11-21 · source benchlm:models |
| 27 | [Gemini 3.8 Flash](https://modelchoice.dev/models/gemini-3-8-flash) | Google | 0.0 | Observed 2026-11-21 · source benchlm:models |
| 27 | [Mistral Large 4](https://modelchoice.dev/models/mistral-large-4) | Mistral | 0.0 | Observed 2026-11-21 · source benchlm:models |

Rows are ordered by the value the source published, highest first. The source does not state whether a higher value is the better result, so this document does not either: for a benchmark that measures a rate of failure, read the table from the bottom.
