TaxPraBen - TaxCalc: leaderboard
TaxPraBen (2026) is a Chinese tax-practice benchmark spanning tax knowledge memorization, understanding and application, evaluated zero-shot. This board is exact-match accuracy in percent on tax payment calculations; higher is better.
Metric: Exact-match accuracy (%) on TaxCalc (tax payment calculation, exact match, 500 items) in TaxPraBen, zero-shot, Chinese prompts written and rated by tax-domain annotators; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 19 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 3 | 19 | #614 |
| 2 | ERNIE-3.5 | 7.6 | |
| 3 | GPT-4o | 7 | #754 |
| 4 | TaxPraBen GLM-4-9B (checkpoint unspecified) | 5.8 | |
| 5 | GPT-3.5 Turbo | 4.6 | #3076 |
| 6 | internlm2.5-7B (TaxPraBen checkpoint unspecified) | 4.2 | |
| 7 | Qwen 2.5 7B (TaxPraBen checkpoint unspecified) | 3.6 | |
| 8 | DeepSeek-R1-Distill-Qwen-7B | 2.4 | #7976 |
| 9 | Yi-6B (TaxPraBen checkpoint unspecified) | 2.4 | |
| 10 | chatglm3-6B | 1.8 | #8290 |
| 11 | Atom (TaxPraBen checkpoint unspecified) | 0.6 | |
| 12 | TaxPraBen DeepSeekllm (checkpoint unspecified) | 0.4 | |
| 13 | Bayling2 | 0.4 | |
| 14 | YAYI2-30B | 0.2 | |
| 15 | LLaMA3 (TaxPraBen checkpoint unspecified) | 0.2 |
No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=taxpraben-taxcalc · How It Works · Data refreshed daily, snapshot 2026-11-22.