Benchmarks · reasoning
Terminal-Bench-Science-AA
Terminal-Bench-Science run by AA: 70 expert-curated scientific research workflows across five domains in a terminal; a task passes only if every one of its tests passes. Pass@1 over three repeats.
36 models · 91 results · data as of 2026-11-03 · artificialanalysis.ai ↗
Leaderboard
Gemini 4 Carbon high
More from this benchmark
9 published measuresScore vs any published measure9 measures: sub-scores, splits, counts
Breakdown
darker is better within each column · click a column to sort| Model | |||||
|---|---|---|---|---|---|
| Gemini 4 Carbon high | 0.614 | 0.549 | 0.804 | 0.667 | 0.583 |
| GPT-6 Astra max | 0.579 | 0.549 | 0.784 | 0.63 | 0.625 |
| Opus 5.5 xhigh | 0.456 | 0.667 | 0.706 | 0.667 | 0.667 |
| GPT-6.1 Sol max | 0.421 | 0.667 | 0.725 | 0.444 | 0.625 |
| GPT-6.1 Terra max | 0.491 | 0.608 | 0.725 | 0.407 | 0.583 |
| Sonnet 5.5 max | 0.509 | 0.569 | 0.647 | 0.519 | 0.292 |
| Gemini 4 Argon high | 0.509 | 0.471 | 0.706 | 0.444 | 0.417 |
| Fable 5.1 max | 0.316 | 0.51 | 0.51 | 0.444 | 0.375 |
| GPT-6 Sol max | 0.228 | 0.333 | 0.314 | 0.296 | 0.375 |
| Opus 5 max | 0.333 | 0.333 | 0.176 | 0.296 | 0.292 |
| GPT-5.6 Sol max | 0.281 | 0.137 | 0.235 | 0.296 | 0.167 |
| Haiku 5.5 max | 0.123 | 0.275 | 0.137 | 0.259 | 0.292 |
| GPT-5.6 Terra max | 0.123 | 0.157 | 0.118 | 0.185 | 0.125 |
| DeepSeek V4.1 Flash 1017 max | 0.123 | 0.098 | 0.0784 | 0.259 | 0.167 |
| Grok 4.7 xhigh | 0.0877 | 0.137 | 0.0588 | 0.0741 | 0.292 |
| Muse Spark 1.3 max | 0.175 | 0.0784 | 0.0784 | 0.0741 | 0.125 |
| Gemini 3.8 Flash medium | 0.158 | 0.137 | 0 | 0.148 | 0.0833 |
| GLM 5.3 max | 0.0702 | 0.196 | 0.0392 | 0.111 | 0.0417 |
| DeepSeek V4.1 Flash max | 0.14 | 0.0588 | 0.098 | 0.0741 | 0.0417 |
| GPT-6 Luna max | 0.0702 | 0.137 | 0.0784 | 0.111 | 0 |
| Kimi K3 max | 0.0526 | 0.098 | 0.0784 | 0 | 0.208 |
| Grok 4.6 high | 0.0702 | 0.0588 | 0 | 0.185 | 0.0417 |
| MiMo V2.6 Pro | 0.0877 | 0.0588 | 0 | 0 | 0.167 |
| GLM 5.3 Flash max | 0.0175 | 0.098 | 0.0196 | 0.0741 | 0.0417 |
| MiMo V2.6 Flash | 0.0175 | 0.098 | 0 | 0.111 | 0 |
All results
91 rows| # | Model | Effort | Score | Subscription $ / task | API $ / task | Harness |
|---|---|---|---|---|---|---|
| 1 | Gemini 4 Carbon | high | 64.8% | $7.12 | $20 | Stirrup |
| 2 | GPT-6 Astra | max | 63.3% | $0.75 | $17 | Stirrup |
| 3 | Gemini 4 Carbon | medium | 62.4% | $6.26 | $18 | Stirrup |
| 4 | Opus 5.5 | xhigh | 61.9% | $0.21 | $14 | Stirrup |
| 5 | Gemini 4 Carbon | low | 61.9% | Stirrup | ||
| 6 | GPT-6 Astra | high | 59.5% | $0.37 | $8.66 | Stirrup |
| 7 | Opus 5.5 | max | 59.0% | $0.30 | $20 | Stirrup |
| 8 | GPT-6.1 Sol | max | 58.1% | $0.19 | $3.34 | Stirrup |
| 9 | GPT-6.1 Terra | max | 57.6% | $2.87 | $2.87 | Stirrup |
| 10 | GPT-6 Astra | xhigh | 54.8% | $0.34 | $7.91 | Stirrup |
| 11 | Sonnet 5.5 | max | 53.3% | $0.36 | $24 | Stirrup |
| 12 | GPT-6 Astra | medium | 53.3% | $0.31 | $7.20 | Stirrup |
| 13 | Gemini 4 Argon | high | 52.9% | $6.66 | $19 | Stirrup |
| 14 | Sonnet 5.5 | xhigh | 52.4% | $0.19 | $13 | Stirrup |
| 15 | GPT-6.1 Sol | xhigh | 50.5% | $0.13 | $2.30 | Stirrup |
| 16 | Opus 5.5 | high | 49.0% | $0.15 | $10 | Stirrup |
| 17 | GPT-6.1 Sol | high | 46.2% | $0.094 | $1.61 | Stirrup |
| 18 | Fable 5.1 | max | 43.3% | $4.38 | $28 | Stirrup |
| 19 | Opus 5.5 | medium | 43.3% | $0.10 | $6.80 | Stirrup |
| 20 | GPT-6.1 Terra | xhigh | 41.9% | $1.04 | $1.04 | Stirrup |
| 21 | Fable 5.1 | xhigh | 40.5% | $4.02 | $26 | Stirrup |
| 22 | Fable 5.1 | high | 39.5% | $3.17 | $20 | Stirrup |
| 23 | GPT-6 Astra | low | 37.1% | $0.18 | $4.12 | Stirrup |
| 24 | GPT-6.1 Sol | medium | 37.1% | $0.051 | $0.87 | Stirrup |
| 25 | Sonnet 5.5 | high | 34.3% | $0.075 | $4.98 | Stirrup |
| 26 | GPT-6.1 Terra | high | 33.3% | $0.62 | $0.62 | Stirrup |
| 27 | GPT-6 Sol | max | 30.0% | Stirrup | ||
| 28 | Fable 5.1 | medium | 29.0% | $2.39 | $15 | Stirrup |
| 29 | Opus 5 | max | 28.6% | Stirrup | ||
| 30 | GPT-6.1 Terra | medium | 24.8% | $0.38 | $0.38 | Stirrup |