FinEval: leaderboard
Chinese financial-domain benchmark covering financial academic knowledge, industry knowledge, security, financial agents, multimodal finance tasks, and rigor testing.
Metric: Weighted average score (%, zero-shot). Source: github.com. Saturation forecast: Estimated already saturated. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3.5 Sonnet | 72.9 | #350 |
| 2 | GPT-4o | 71.9 | #343 |
| 3 | Qwen 2.5 72B Instruct | 69.4 | #454 |
| 4 | Gemini 1.5 Pro | 69.2 | #445 |
| 5 | GPT-4o Mini | 66.2 | #599 |
| 6 | Gemini 1.5 Flash | 65.6 | #618 |
| 7 | Qwen 2.5 7B Instruct | 62.3 | #862 |
| 8 | Yi 1.5 34B Chat | 61.5 | #800 |
| 9 | internlm2.5-20B-chat | 58.9 | #793 |
| 10 | GLM-4 9B Chat | 58.4 | #916 |
| 11 | internlm2-chat-20B | 58 | #999 |
| 12 | Yi-1.5-9B Chat | 56.9 | #1031 |
| 13 | Baichuan2-13B-Chat | 47.8 | #1255 |
| 14 | chatglm3-6B | 43.2 | #1439 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=fineval · How It Works · Data refreshed daily, snapshot 2026-10-20.