HiThink Research, 2025
BizFinBench
Chinese business-finance queries drawn from real use, in five areas: numerical calculation, reasoning, information extraction, prediction recognition and knowledge QA (nine categories).
About this benchmark
- Size
- 6,781 questions, 9 categories
- Public split
- All 6,781 questions on Hugging Face (test split per subset)
- Grading
- Accuracy for six tasks, with rule-based answer extraction. GPT-4o judge scores the other three (tool use, knowledge QA, data description). The paper proposes the IteraJudge framework.
- Licence
- CC-BY-NC-4.0 (data, research use only); Apache-2.0 (code)
- Submission
- maker-only
- Products allowed
- No
- Latest version
- v1 (paper 2025-05-26), marked outdated by the maker; BizFinBench.v2 released 2026-01-09 (29,578 questions, ICML 2026)
- What it does not cover
- Chinese only in v1. No live retrieval, no agentic tool loops beyond one tool-usage task, and no English filings.
Results
10 modelsEach score shows who reported it: the benchmark's authors (quoted from their paper), the benchmark's official board, or Moneda (run with the benchmark's own grader and rules). We do not copy other aggregators.
BizFinBench v1 (paper 2025-05-26), marked outdated by the maker; BizFinBench.v2 released 2026-01-09 (29,578 questions, ICML 2026)
average score across 9 tasks, higher is better. 10 models.
Updated 26 May 2025
- ChatGPT-o373.9
- DeepSeek-R1 (671B)73.0
- GPT-4o71.8
- DeepSeek-V3 (671B)71.6
- ChatGPT-o4-mini71.3
- Gemini-2.0-Flash69.8
- Qwen2.5-VL-32B68.6
- Qwen3-32B68.3
- Qwen2.5-72B-Instruct67.7
- Qwen2.5-VL-72B67.5
BizFinBench v1 (paper 2025-05-26), marked outdated by the maker; BizFinBench.v2 released 2026-01-09 (29,578 questions, ICML 2026)
average score across 9 tasks, higher is better. 10 models.
Updated 26 May 2025
- 1ChatGPT-o373.9
- 2DeepSeek-R1 (671B)73.0
- 3GPT-4o71.8
- 4DeepSeek-V3 (671B)71.6
- 5ChatGPT-o4-mini71.3
- 6Gemini-2.0-Flash69.8
- 7Qwen2.5-VL-32B68.6
- 8Qwen3-32B68.3
- 9Qwen2.5-72B-Instruct67.7
- 10Qwen2.5-VL-72B67.5
| 1 | ChatGPT-o3 | OpenAI | 73.9 | Default | Paper Table 3 Reported by the authors, 26 May 2025 |
|---|---|---|---|---|---|
| 2 | DeepSeek-R1 (671B) | DeepSeek | 73.0 | Default | Paper Table 3 Reported by the authors, 26 May 2025 |
| 3 | GPT-4o | OpenAI | 71.8 | Default | Paper Table 3 Reported by the authors, 26 May 2025 |
| 4 | DeepSeek-V3 (671B) | DeepSeek | 71.6 | Default | Paper Table 3 Reported by the authors, 26 May 2025 |
| 5 | ChatGPT-o4-mini | OpenAI | 71.3 | Default | Paper Table 3 Reported by the authors, 26 May 2025 |
| 6 | Gemini-2.0-Flash | 69.8 | Default | Paper Table 3 Reported by the authors, 26 May 2025 | |
| 7 | Qwen2.5-VL-32B | Qwen | 68.6 | Default | Paper Table 3 Reported by the authors, 26 May 2025 |
| 8 | Qwen3-32B | Qwen | 68.3 | Default | Paper Table 3 Reported by the authors, 26 May 2025 |
| 9 | Qwen2.5-72B-Instruct | Qwen | 67.7 | Default | Paper Table 3 Reported by the authors, 26 May 2025 |
| 10 | Qwen2.5-VL-72B | Qwen | 67.5 | Default | Paper Table 3 Reported by the authors, 26 May 2025 |
Papers and links
Last checked 1 Nov 2026. Something out of date or wrong? Send a correction.