FinEval - Financial Agent: leaderboard
FinEval's financial agent section: chain-of-thought planning, retrieval-augmented answering, financial tasks, multi-turn dialogue, multi-document question answering and API selection and invocation. Responses are scored by similarity (%) to reference answers under zero-shot prompting; higher is better.
Metric: Similarity 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 | 79.3 | #747 |
| 2 | GPT-4o | 73.9 | #741 |
| 3 | GPT-4o Mini | 72.9 | #1627 |
| 4 | Gemini 1.5 Pro | 72.8 | #1053 |
| 5 | Gemini 1.5 Flash | 70.9 | #1671 |
| 6 | Qwen 2.5 72B Instruct | 68.4 | #1093 |
| 7 | Qwen 2.5 7B Instruct | 66.7 | #2583 |
| 8 | Yi 1.5 34B Chat | 66 | #2513 |
| 9 | internlm2.5-20B-chat | 63.1 | #2563 |
| 10 | Yi-1.5-9B Chat | 61.1 | #4060 |
| 11 | internlm2-chat-20B | 60.9 | #2896 |
| 12 | GLM-4 9B Chat | 60.2 | #2782 |
| 13 | Baichuan2-13B-Chat | 55.7 | #6615 |
| 14 | chatglm3-6B | 49.6 | #8269 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=fineval-financial-agent · How It Works · Data refreshed daily, snapshot 2026-11-12.