TaxEval
As of Nov 3, 2026, TaxEval (v2) 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 Oct 27, 2026.
- Source
- Vals AI · this benchmark on Vals AI
- Grading
- AI judge (LLM-graded rubric) — Answer correctness and stepwise reasoning are graded by an LLM judge.
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
- Muse Spark 1.2 · 80.376 · Oct 27, 2026
- Muse Spark 1.1 · 79.722 · Oct 27, 2026
- Muse Spark · 77.678 · Oct 27, 2026
- Claude Sonnet 4.6 · 77.106 · Oct 27, 2026
- Claude Fable 5 · 76.942 · Oct 27, 2026
| Model | Score | Date | Kind |
|---|---|---|---|
| Muse Spark 1.2 | 80.376 | 2026-10-27 | snapshot |
| Muse Spark 1.1 | 79.722 | 2026-10-27 | snapshot |
| Muse Spark | 77.678 | 2026-10-27 | snapshot |
| Claude Sonnet 4.6 | 77.106 | 2026-10-27 | snapshot |
| Claude Fable 5 | 76.942 | 2026-10-27 | snapshot |
| GPT-5.6 Terra | 76.166 | 2026-10-27 | snapshot |
| GPT-5.6 Luna | 76.166 | 2026-10-27 | snapshot |
| Claude Opus 4.6 Thinking | 75.961 | 2026-10-27 | snapshot |
| Claude Fable 5.1 | 75.961 | 2026-10-27 | snapshot |
| Grok 3 | 75.879 | 2026-10-27 | snapshot |
| GPT-5.2 | 75.756 | 2026-10-27 | snapshot |
| Kimi K3 | 75.716 | 2026-10-27 | snapshot |
| Grok 4 Fast Reasoning | 75.697 | 2026-10-27 | snapshot |
| Claude Opus 4.8 | 75.634 | 2026-10-27 | snapshot |
| Claude Sonnet 5 | 75.634 | 2026-10-27 | snapshot |
| GLM 5.3 Flash | 75.593 | 2026-10-27 | snapshot |
| Qwen3 8 Max | 75.552 | 2026-10-27 | snapshot |
| Inkling Small | 75.511 | 2026-10-27 | snapshot |
| Qwen3 7 Max | 75.306 | 2026-10-27 | snapshot |
| Inkling | 75.306 | 2026-10-27 | snapshot |
| Claude Opus 4.7 | 75.266 | 2026-10-27 | snapshot |
| GPT-5 Mini | 75.225 | 2026-10-27 | snapshot |
| Claude Opus 5 | 75.143 | 2026-10-27 | snapshot |
| GPT-4.1 | 75.061 | 2026-10-27 | snapshot |
| GPT-5.5 | 74.98 | 2026-10-27 | snapshot |
| Gemini 3.6 Flash | 74.857 | 2026-10-27 | snapshot |
| GPT-5.1 | 74.857 | 2026-10-27 | snapshot |
| Claude Opus 4.5 20251101 Thinking | 74.856 | 2026-10-27 | snapshot |
| o4-mini | 74.776 | 2026-10-27 | snapshot |
| GPT-5.6 Sol | 74.775 | 2026-10-27 | snapshot |