Vals AI Terminal-Bench 2.0 — leaderboard
Metric: Accuracy (%). Source: www.vals.ai. 67 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 73.2 |
| 2 | Claude Opus 4.8 | 70.04 |
| 3 | Claude Opus 4.7 | 68.54 |
| 4 | Gemini 3.1 Pro (Preview) | 67.42 |
| 5 | Gemini 3.5 Flash | 67.42 |
| 6 | GPT-5.3 Codex | 64.05 |
| 7 | Claude Sonnet 4.6 | 59.55 |
| 8 | Muse Spark | 59.55 |
| 9 | Qwen 3.7 Max | 59.18 |
| 10 | Claude Opus 4.6 (Thinking) | 58.43 |
| 11 | Claude Opus 4.5 (20251101) | 58.43 |
| 12 | GPT-5.4 (2026-03-05) | 58.43 |
| 13 | Kimi K2.6 | 57.3 |
| 14 | DeepSeek V4 Pro | 56.18 |
| 15 | Gemini 3 Pro (Preview) | 55.06 |
Interactive version: theaggregate.ai/benchmark?slug=vals-ai-terminal-bench-2-0 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.