Terminal-Bench 3.0: leaderboard
Terminal-Bench 3.0 measures whether agents can complete difficult, economically valuable tasks in command-line environments. The evaluated model works from text inputs in terminal settings. The pinned metric is Pass@1. Source: BenchmarkList’s attributed result table.
Metric: Pass@1 (%). Source: benchmarklist.com. Saturation forecast: Around October 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 (Max) | 42.7 | #16 (Claude Opus 5) |
| 2 | GPT-5.6 Sol (Max) | 34.59 | #22 (GPT-5.6 Sol) |
| 3 | Claude Fable 5 (Max) | 34.05 | #20 (Claude Fable 5) |
| 4 | GLM-5.3 (Max) | 32.43 | #44 (GLM-5.3) |
| 5 | Grok 4.6 (High) | 26.49 | #29 (Grok 4.6) |
| 6 | Claude Opus 4.8 (Max) | 21.08 | #39 (Claude Opus 4.8) |
| 7 | GPT-5.6 Terra (Max) | 20.81 | #40 (GPT-5.6 Terra) |
| 8 | Claude Sonnet 5 (Max) | 14.59 | #65 (Claude Sonnet 5) |
| 9 | GPT-5.6 Luna (Max) | 14.32 | #79 (GPT-5.6 Luna) |
| 10 | GLM-5.2 (Max) | 4.59 | #80 (GLM-5.2) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=terminal-bench-3-0 · How It Works · Data refreshed daily, snapshot 2026-11-05.