AA Terminal-Bench Science 0.1 - Life Sciences: leaderboard
Artificial Analysis pass@1 on the 19 life 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 May 2027. 100 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Carbon (High) | 61.4 | #2 (Gemini 4 Carbon) |
| 2 | GPT-6 Astra (Max) | 57.89 | #5 (GPT-6 Astra) |
| 3 | GPT-6 Astra (High) | 57.89 | #5 (GPT-6 Astra) |
| 4 | Gemini 4 Argon (High) | 50.88 | #9 (Gemini 4 Argon) |
| 5 | GPT-6 Astra (xHigh) | 49.12 | #5 (GPT-6 Astra) |
| 6 | GPT-6 Astra (Medium) | 49.12 | #5 (GPT-6 Astra) |
| 7 | GPT-6.1 Terra (Max) | 49.12 | #13 (GPT-6.1 Terra) |
| 8 | Kimi K3.1 Pro | 43.86 | #16 |
| 9 | GPT-6.1 Sol (Max) | 42.11 | #7 (GPT-6.1 Sol) |
| 10 | Muse Wonder Full (Max) | 38.6 | #11 (Muse Wonder Full) |
| 11 | GPT-6.1 Sol (xHigh) | 36.84 | #7 (GPT-6.1 Sol) |
| 12 | GPT-6.1 Sol (High) | 35.09 | #7 (GPT-6.1 Sol) |
| 13 | GPT-6.1 Terra (xHigh) | 35.09 | #13 (GPT-6.1 Terra) |
| 14 | Claude Opus 5 (Max) | 33.33 | #15 (Claude Opus 5) |
| 15 | GPT-6 Astra (Low) | 33.33 | #5 (GPT-6 Astra) |
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-life-sciences · How It Works · Data refreshed daily, snapshot 2026-10-29.