FinEval - Multimodal: leaderboard
Metric: Weighted average accuracy (%). Source: github.com. Saturation forecast: Estimated already saturated. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | InternVL3-78B | 72.5 |
| 2 | InternVL2.5-78B | 71.5 |
| 3 | GPT-4o (2024-11-20) | 68.5 |
| 4 | InternVL3-8B | 65.4 |
| 5 | Qwen 2.5 VL 7B | 65.4 |
| 6 | Gemini 2.5 Pro (Preview 03-25) | 64.7 |
| 7 | Claude 3.7 Sonnet (20250219) | 62.9 |
| 8 | MiniCPM-V-2.6 | 60.1 |
| 9 | Llama 3.2 11B Instruct | 50.9 |
Interactive version: theaggregate.ai/benchmark?slug=fineval-multimodal · How It Works · Data refreshed daily, snapshot 2026-09-26.