Multimodal Chart Reasoning on FinChart (official split)
97.86TF Scoreo3
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| o3Backbone=o32026.06 | 97.86 | 92.17 | 60.79 | 83.89 | |
| o4 miniBackbone=o4 mini2026.06 | 97.61 | 91.96 | 60.18 | 83.53 | |
| Gemini 2.5 ProBackbone=Gemini 2.5 Pro2026.06 | 97.27 | 89.7 | 63.46 | 83.73 | |
| MOCA-AGENTBackbone=Qwen3.6-27B + VLM2026.06 | 97.27 | 84.85 | 74.65 | 85.59 | |
| Claude Sonnet 4Backbone=Claude Sonnet 42026.06 | 96.94 | 91.66 | 63.59 | 84.32 | |
| GPT 4.1Backbone=GPT 4.12026.06 | 96.18 | 83.19 | 62.54 | 80.88 | |
| Qwen2.5 VLBackbone=Qwen2.5 VL (7B)2026.06 | 95.09 | 82.81 | 37.29 | 72.16 | |
| GPT 4oBackbone=GPT 4o2026.06 | 93.92 | 77.57 | 57.59 | 76.62 | |
| Mistral 3.1Backbone=Mistral 3.1 (24B)2026.06 | 92.95 | 84.17 | 44.9 | 74.37 | |
| Qwen2 VLBackbone=Qwen2 VL (7B)2026.06 | 91.86 | 19.4 | 24.77 | 45.76 | |
| GPT 4.1 miniBackbone=GPT 4.1 mini2026.06 | 91.61 | 86.77 | 60.3 | 79.8 | |
| LLaMa 4Backbone=LLaMa 4 (17B)2026.06 | 89.56 | 59.7 | 59.78 | 69.89 | |
| Gemma 3Backbone=Gemma 3 (12B)2026.06 | 89.14 | 70.94 | 27.22 | 62.89 | |
| Sa2VABackbone=Sa2VA (8B)2026.06 | 86.62 | 75.79 | 19.43 | 61.12 | |
| CosmosBackbone=Cosmos (7B)2026.06 | 70.93 | 43.49 | 20.18 | 45.22 | |
| LLaMa 3.2Backbone=LLaMa 3.2 (11B)2026.06 | 63.46 | 64 | 5.12 | 44.65 | |
| LLaVa v1.6Backbone=LLaVa v1.6 (7B)2026.06 | 63.3 | 36.94 | 1.84 | 34.47 | |
| DeepSeek VL2Backbone=DeepSeek VL2 (2.8B)2026.06 | 57.97 | 6.77 | 4.73 | 23.5 |