terminal-bench@2.1
terminal-bench@2.1 is run by Terminal-Bench. It has ranked 17 models, of which the catalogue holds 13, scoring from 58.7 to 83.8 on Accuracy.
measured by Terminal-Bench · the board itself
| Place | Model | Metric | Score |
|---|---|---|---|
| 1stof 17 | Claude Fable 5 | Accuracy (xhigh) | 83.8 |
| 2ndof 17 | GPT-5.5 | Accuracy (xhigh) | 83.1 |
| 3rdof 17 | Claude Fable 5 | Accuracy (high) | 80.4 |
| 4thof 17 | Grok 4.5 | Accuracy (high) | 79.3 |
| 5thof 17 | Claude Opus 4.8 | Accuracy (high) | 78.9 |
| 6thof 17 | GPT-5.6 Terra | Accuracy (max) | 78.4 |
| 8thof 17 | Muse Spark 1.1 | Accuracy (xhigh) | 76.2 |
| 9thof 17 | GPT-5.6 Luna | Accuracy (max) | 75.7 |
| 10thof 17 | Claude Sonnet 5 | Accuracy (high) | 74.6 |
| 11thof 17 | gemini-3-pro-preview | Accuracy | 73.9 |
| 12thof 17 | Claude Opus 4.7 | Accuracy | 68.9 |
| 14thof 17 | Gemini 3.1 Pro Preview | Accuracy | 65.8 |
| 17thof 17 | GLM 5.1 | Accuracy | 58.7 |
Read from the board on 2026-08-25