Tax Agent Bench — Forms & Filings All-Pass
Professional Work · 2026-09-23
All-rubric-item pass rate on the 30 forms-and-filings questions.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5 | 33.3 |
| Muse Spark 1.2 | 33.3 |
| Claude Fable 5.1 | 30.0 |
| Grok 4.7 | 30.0 |
| Claude Fable 5 | 26.7 |
| Gemini 4 Argon | 26.7 |
| Hy4 Preview | 26.7 |
| Mistral Large 4 Preview | 26.7 |
| Muse Spark 1.3 | 26.7 |
| GLM-5.3 Flash | 23.3 |
| Grok 4.6 | 23.3 |
| Sonnet 5 | 20.0 |
| Ember-1 | 20.0 |
| Gemini 3.8 Flash | 20.0 |
| GLM-5.3 | 20.0 |
| GPT-5.6 Sol | 20.0 |
| Kimi K3 | 20.0 |
| MiMo-V2.6-Flash | 20.0 |
| MiMo-V2.6-Pro | 20.0 |
| MiniMax M3 | 20.0 |
| DeepSeek-V4-Flash-0731 | 16.7 |
| Gemini 3.5 Flash | 16.7 |
| GPT-5.6 Terra | 16.7 |
| Grok 4.5 | 16.7 |
| Qwen3.8 27B | 16.7 |
| Step 5 Preview | 16.7 |
| Opus 4.8 | 13.3 |
| DeepSeek V4.1 Flash | 13.3 |
| DeepSeek V4 Pro 0813 | 13.3 |
| GPT-5.6 Luna | 13.3 |
| GPT-6 Luna | 13.3 |
| Inkling-Small | 13.3 |
| Gemini 3.7 Flash | 10.0 |
| GPT-5.5 | 10.0 |
| Qwen3.7-Max (source unspecified) | 10.0 |
| Opus 4.7 | 6.7 |
| Sonnet 4.6 | 6.7 |
| Gemini 3.6 Flash | 6.7 |
| Gemini 3 Flash Preview | 6.7 |
| GPT-6 Astra | 6.7 |