τ³-Banking Leaderboard
τ³-Banking Leaderboard is run by Sierra. It has ranked 28 models, of which the catalogue holds 26, scoring from 9.8 to 55.2 on Pass^1.
measured by Sierra · the board itself
| Place | Model | Metric | Score |
|---|---|---|---|
| 1stof 28 | Qwen 3.8 Max | Pass^1 | 55.2 |
| 2ndof 28 | Claude Opus 5 | Pass^1 | 48.7 |
| 3rdof 28 | Grok 4.5 | Pass^1 | 47.9 |
| 4thof 28 | GPT-5.6 Sol | Pass^1 | 46.9 |
| 5thof 28 | GPT-5.5 | Pass^1 | 44.6 |
| 6thof 28 | Muse Spark 1.1 | Pass^1 | 40.5 |
| 7thof 28 | Claude Opus 4.7 | Pass^1 | 40.2 |
| 8thof 28 | Claude Opus 4.8 | Pass^1 | 39.7 |
| 9thof 28 | Claude Fable 5 | Pass^1 | 39.7 |
| 10thof 28 | GPT-5.4 | Pass^1 | 39.4 |
| 11thof 28 | Kimi K3 | Pass^1 | 37.1 |
| 12thof 28 | GLM 5.2 | Pass^1 | 37.1 |
| 13thof 28 | GPT-5.2 | Pass^1 | 32.2 |
| 15thof 28 | Gemini 3 Flash Preview | Pass^1 | 27.3 |
| 16thof 28 | Claude Opus 4.6 | Pass^1 | 27.3 |
| 17thof 28 | Gemini 3.1 Pro Preview | Pass^1 | 26 |
| 18thof 28 | Claude Sonnet 4.5 | Pass^1 | 25.3 |
| 19thof 28 | Inkling | Pass^1 | 25 |
| 20thof 28 | Claude Opus 4.5 | Pass^1 | 24.7 |
| 21stof 28 | gemini-3-pro-preview | Pass^1 | 18 |
| 22ndof 28 | Grok 4.20 | Pass^1 | 18 |
| 23rdof 28 | Grok 4 Fast | Pass^1 | 15.7 |
| 24thof 28 | Gemini 2.5 Pro | Pass^1 | 13.7 |
| 25thof 28 | Grok 4.1 Fast | Pass^1 | 13.1 |
| 27thof 28 | GLM-5 | Pass^1 | 9.8 |
| 28thof 28 | Qwen3.5 397B A17B | Pass^1 | 9.8 |
Read from the board on 2026-08-25