Tax Agent Bench
As of Nov 3, 2026, Tax Agent Bench (v1) is maintained by Vals AI. Metric accuracy (%). Ceiling 100. Saturation unknown. Contamination risk low.
- Maintainer
- Vals AI
- Metric
- accuracy (%)
- Ceiling
- 100
- Saturation
- unknown
- Contamination
- low
- Cadence
- Declared cadence: as Vals updates (checked daily). Catalog snapshots: 1, latest Nov 2, 2026.
- Source
- Vals AI · this benchmark on Vals AI
- Grading
- AI judge (LLM-graded rubric) — Research-grade corporate tax questions graded by an LLM judge against expert rubrics.
Source: Vals AI. A proprietary private test set run by Vals; scores are republished with permission. Rows are dated the day isitsota first saw them, and only the benchmark's standard harness counts.
Categories
Where it counts
- Finance — secondary, weight 0.35
Top 5
Current top scores
- Gemini 4 Carbon · 51.785 · Nov 2, 2026
- Claude Fable 5.1 · 49.194 · Nov 2, 2026
- Claude Opus 5.5 · 45.573 · Nov 2, 2026
- Claude Opus 5 · 45.548 · Nov 2, 2026
- Gemini 4 Argon · 44.56 · Nov 2, 2026
| Model | Score | Date | Kind |
|---|---|---|---|
| Claude Fable 5.1 | 49.194 | 2026-11-02 | snapshot |
| Claude Opus 5.5 | 45.573 | 2026-11-02 | snapshot |
| Claude Opus 5 | 45.548 | 2026-11-02 | snapshot |
| Gemini 4 Argon | 44.56 | 2026-11-02 | snapshot |
| DeepSeek V4.1 Flash 1017 | 44.35 | 2026-11-02 | snapshot |
| Muse Wonder Full | 42.997 | 2026-11-02 | snapshot |
| Muse Spark 1.3 | 42.435 | 2026-11-02 | snapshot |
| Claude Sonnet 5.5 | 41.959 | 2026-11-02 | snapshot |
| GLM 5.3 | 39.698 | 2026-11-02 | snapshot |
| Claude Fable 5 | 38.281 | 2026-11-02 | snapshot |
| Muse Wonder | 36.269 | 2026-11-02 | snapshot |
| Muse Spark 1.3 | 35.974 | 2026-11-02 | snapshot |
| Grok 4.7 | 34.197 | 2026-11-02 | snapshot |
| Kimi K3 | 33.368 | 2026-11-02 | snapshot |
| Grok 4.6 | 33.141 | 2026-11-02 | snapshot |
| Muse Spark 1.2 | 32.798 | 2026-11-02 | snapshot |
| Hy4 Preview | 32.617 | 2026-11-02 | snapshot |
| Mimo V2.6 Pro | 32.388 | 2026-11-02 | snapshot |
| MiniMax M3.1 Flash | 32.136 | 2026-11-02 | snapshot |
| Gemini 3.8 Flash | 32.124 | 2026-11-02 | snapshot |
| Qwen3 8 Max | 32.095 | 2026-11-02 | snapshot |
| Claude Haiku 5.5 | 31.922 | 2026-11-02 | snapshot |
| GLM 5.3 Flash | 31.55 | 2026-11-02 | snapshot |
| GPT-5.6 Sol | 31.088 | 2026-11-02 | snapshot |
| Qwen3.8 27B | 30.894 | 2026-11-02 | snapshot |
| Mistral Large 4 | 30.391 | 2026-11-02 | snapshot |
| GPT-5.6 Terra | 30.035 | 2026-11-02 | snapshot |
| Claude Opus 4.8 | 29.999 | 2026-11-02 | snapshot |
| Claude Sonnet 5 | 29.374 | 2026-11-02 | snapshot |
| Mimo V2.6 Flash | 28.507 | 2026-11-02 | snapshot |