Terminal Bench: leaderboard
Terminal and command-line interaction tasks for evaluating agent performance.
Metric: Score (imported). Source: neosignal.io. Source: benchmarklist.com. Saturation forecast: Around February 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 | 64.9 | #110 |
| 2 | Gemini 3 Pro | 64.3 | #80 |
| 3 | Claude Opus 4.5 | 63.1 | #83 |
| 4 | Kimi K2 (Thinking) | 35.7 | #243 (Kimi K2) |
| 5 | Gemini 2.5 Pro (Preview 06-05) | 32.6 | #106 |
| 6 | Grok 4 | 27.2 | #181 |
| 7 | Grok Code Fast 1 | 25.8 | #337 |
| 8 | Qwen 3 Max Instruct | 25.4 | #259 |
| 9 | GPT-OSS-120B | 18.7 | #341 |
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 · How It Works · Data refreshed daily, snapshot 2026-10-20.