Tax Agent Bench — Fact-Pattern All-Pass
Professional Work · 2026-09-23
All-rubric-item pass rate on the 43 fact-pattern questions.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Fable 5.1 | 51.2 |
| Claude Opus 5 | 48.8 |
| Claude Opus 5.5 | 46.5 |
| Muse Spark 1.3 | 46.5 |
| MiMo-V2.6-Flash | 41.9 |
| GPT-5.6 Terra | 37.2 |
| Gemini 3.8 Flash | 34.9 |
| GLM-5.3 | 34.9 |
| Grok 4.7 | 34.9 |
| Kimi K3 | 32.6 |
| DeepSeek V4 Pro 0813 | 30.2 |
| GPT-6 Luna | 30.2 |
| Hy4 Preview | 30.2 |
| MiMo-V2.6-Pro | 30.2 |
| Sonnet 5 | 27.9 |
| DeepSeek V4.1 Flash | 27.9 |
| GPT-5.5 | 27.9 |
| GPT-6 Astra | 27.9 |
| Grok 4.6 | 27.9 |
| Inkling | 27.9 |
| GPT-5.6 Luna | 25.6 |
| GPT-5.6 Sol | 23.3 |
| Gemini 3.7 Flash | 18.6 |
| GPT-6 Sol | 16.3 |
| MiniMax M3 | 16.3 |
| Mercury 2.5 | 2.3 |