Terminal-Bench 2.0: leaderboard
89 terminal/CLI tasks testing model ability to use command-line tools, navigate filesystems, and automate workflows. Scored by accuracy across agent-model pairings.
Metric: Accuracy (%). Source: www.tbench.ai. Saturation forecast: Around January 2027. 59 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 84.72 | #31 |
| 2 | GPT-5.4 (2026-03-05) | 81.8 | #59 |
| 3 | Gemini 3.1 Pro (Preview) | 80.22 | #58 |
| 4 | GPT-5.3 Codex | 78.43 | #69 |
| 5 | Claude Opus 4.6 | 69.9 | #63 |
| 6 | Gemini 3 Pro (Preview) | 69.43 | #66 |
| 7 | GPT-5.2 Codex | 66.52 | #95 |
| 8 | GPT-5.2 | 64.94 | #110 |
| 9 | GPT-5.2 (2025-12-11) (Medium) | 64.9 | |
| 10 | Gemini 3 Flash (Preview) | 64.27 | #82 |
| 11 | Claude Opus 4.5 (20251101) | 63.1 | #74 |
| 12 | GPT-5.1 Codex Mini | 61.57 | #286 |
| 13 | GPT-5.1 Codex Max | 60.45 | #100 |
| 14 | GPT-5.1 Codex | 57.8 | #159 |
| 15 | Grok 4.20 | 57.3 | #175 |
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-2-0 · How It Works · Data refreshed daily, snapshot 2026-10-20.