Benchmarks
Terminal-Bench
Models scored
203
Best score
0.9
Metric
Accuracy mean
Terminal-Bench is an AI evaluation tracked by Epoch AI, with 203 scored model versions on record. The scores here are reported by third parties rather than produced by Epoch AI running the evaluation itself.
Scored models span June 2025 to April 2026. The best score recorded is 0.9, measured as Accuracy mean. Models come from Anthropic, OpenAI, Google DeepMind, xAI, Z.ai (Zhipu AI), MiniMax, Moonshot, DeepSeek and others.
Full record
- Score metric
- Accuracy mean
- Models scored
- 203
- Best score recorded
- 0.9
- Earliest model scored
- Jun 17, 2025
- Most recent model scored
- Apr 23, 2026
- Third-party reported
- Yes
Leaderboard
| 01 | claude-opus-4-7_unknown | Anthropic | 0.9 |
| 02 | gpt-5.5_unknown | OpenAI | 0.85 |
| 03 | gpt-5.4-2026-03-05_unknown | OpenAI | 0.82 |
| 04 | gemini-3.1-pro-preview | Google DeepMind | 0.8 |
| 05 | claude-opus-4-6_unknown | Anthropic | 0.8 |
| 06 | gpt-5.3-codex | OpenAI | 0.78 |
| 07 | claude-opus-4-6 | Anthropic | 0.7 |
| 08 | gemini-3-pro-preview | Google DeepMind | 0.69 |
| 09 | gpt-5.2-codex | OpenAI | 0.67 |
| 10 | gpt-5.2-2025-12-11_unknown | OpenAI | 0.65 |
| 11 | gpt-5.2-2025-12-11_medium | OpenAI | 0.65 |
| 12 | gemini-3-flash-preview | Google DeepMind | 0.64 |
| 13 | claude-opus-4-5-20251101 | Anthropic | 0.63 |
| 14 | claude-opus-4-5-20251101_unknown | Anthropic | 0.63 |
| 15 | gpt-5.1-codex-mini | OpenAI | 0.62 |
| 16 | gpt-5.1-codex-max | OpenAI | 0.6 |
| 17 | claude-opus-4-5-20251101_128K | Anthropic | 0.59 |
| 18 | gpt-5.1-codex | OpenAI | 0.58 |
| 19 | grok-4-20 | xAI | 0.57 |
More benchmarks
SourceEpoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/benchmarks. Licensed under CC BY 4.0.
← All benchmarks