AA Terminal-Bench 2.1: leaderboard
Metric: Pass@1 (%). Source: artificialanalysis.ai. Saturation forecast: Around January 2027. 240 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-6 Astra (High) | 89.89 |
| 2 | GPT-5.6 Sol (xHigh) | 89.51 |
| 3 | GPT-6 Astra (Medium) | 89.51 |
| 4 | Claude Opus 5 (Max) | 89.14 |
| 5 | GPT-6 Astra (xHigh) | 89.14 |
| 6 | Qwen 3.8 Max (0902) | 88.76 |
| 7 | GPT-6 Astra (Max) | 88.39 |
| 8 | Grok 4.6 (High) | 88.39 |
| 9 | GPT-5.6 Sol (Max) | 88.01 |
| 10 | GPT-5.6 Terra (Max) | 88.01 |
| 11 | Grok 4.6 (xHigh) | 88.01 |
| 12 | Claude Opus 5 (xHigh) | 88.01 |
| 13 | GPT-6 Astra (Low) | 88.01 |
| 14 | Gemini 3.8 Flash (High) | 87.64 |
| 15 | Claude Opus 5 (High) | 87.64 |
Interactive version: theaggregate.ai/benchmark?slug=aa-terminal-bench-2-1 · How It Works · Data refreshed daily, snapshot 2026-10-10.