TaxPraBen - TaxTopic: leaderboard
TaxPraBen (2026) is a Chinese tax-practice benchmark spanning tax knowledge memorization, understanding and application, evaluated zero-shot. This board is accuracy in percent on classifying tax texts by topic; higher is better.
Metric: Accuracy (%) on TaxTopic (tax topic classification, 1,000 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 | TaxPraBen GLM-4-9B (checkpoint unspecified) | 63 | |
| 2 | GPT-4o | 55.6 | #751 |
| 3 | GPT-3.5 Turbo | 44.2 | #3005 |
| 4 | internlm2.5-7B (TaxPraBen checkpoint unspecified) | 43.4 | |
| 5 | ERNIE-3.5 | 42.4 | |
| 6 | Grok 3 | 34.6 | #611 |
| 7 | Qwen 2.5 7B (TaxPraBen checkpoint unspecified) | 34 | |
| 8 | DeepSeek-R1-Distill-Qwen-7B | 22.6 | #8055 |
| 9 | LLaMA3 (TaxPraBen checkpoint unspecified) | 22.2 | |
| 10 | Atom (TaxPraBen checkpoint unspecified) | 21.1 | |
| 11 | Yi-6B (TaxPraBen checkpoint unspecified) | 21 | |
| 12 | ChnLLaMA3 (TaxPraBen checkpoint unspecified) | 19.3 | |
| 13 | Mistral-V0.3 (TaxPraBen checkpoint unspecified) | 15.2 | |
| 14 | Baichuan2 (TaxPraBen checkpoint unspecified) | 14.6 | |
| 15 | TaxPraBen DeepSeekllm (checkpoint unspecified) | 12.8 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=taxpraben-taxtopic · How It Works · Data refreshed daily, snapshot 2026-11-21.