AA Terminal-Bench Science 0.1 - Earth Sciences: leaderboard
Artificial Analysis pass@1 on the 8 earth science tasks in Terminal-Bench Science 0.1.0. The mini-swe-agent harness uses a 1,000-step cap, upstream resource/time limits and isolated verifier containers. All tests must pass; scores average three repeats per task and are percentages.
Metric: Pass@1 (%). Source: artificialanalysis.ai. Saturation forecast: Around April 2027. 100 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6.1 Sol (xHigh) | 66.67 | #7 (GPT-6.1 Sol) |
| 2 | GPT-6 Astra (Max) | 62.5 | #5 (GPT-6 Astra) |
| 3 | GPT-6.1 Sol (Max) | 62.5 | #7 (GPT-6.1 Sol) |
| 4 | GPT-6.1 Terra (Max) | 58.33 | #13 (GPT-6.1 Terra) |
| 5 | Gemini 4 Carbon (High) | 58.33 | #2 (Gemini 4 Carbon) |
| 6 | GPT-6 Astra (xHigh) | 54.17 | #5 (GPT-6 Astra) |
| 7 | GPT-6 Astra (High) | 54.17 | #5 (GPT-6 Astra) |
| 8 | GPT-6 Astra (Medium) | 50 | #5 (GPT-6 Astra) |
| 9 | GPT-6.1 Terra (High) | 50 | #13 (GPT-6.1 Terra) |
| 10 | Gemini 4 Argon (High) | 41.67 | #9 (Gemini 4 Argon) |
| 11 | GPT-6 Astra (Low) | 41.67 | #5 (GPT-6 Astra) |
| 12 | GPT-6.1 Sol (Medium) | 41.67 | #7 (GPT-6.1 Sol) |
| 13 | GPT-6 Sol (Max) | 37.5 | #18 (GPT-6 Sol) |
| 14 | GPT-6.1 Sol (High) | 37.5 | #7 (GPT-6.1 Sol) |
| 15 | Muse Wonder Full (Max) | 37.5 | #11 (Muse Wonder Full) |
No result here: #3 Claude Opus 5.5, #10 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=aa-terminal-bench-science-0-1-earth-sciences · How It Works · Data refreshed daily, snapshot 2026-10-29.