Chart Understanding on ChartQA (Score)
88.4ScoreDeepEyesV2
Evaluation Results
| Method | Links | |
|---|---|---|
| DeepEyesV2Tool=General, Param Size=7B2025.11 | 88.4 | |
| LLaDA-oModel Category=Unified Models2026.04 | 87.9 | |
| InternVL3Tool=✗, Param Size=8B2025.11 | 86.6 | |
| Qwen2.5-VLTool=✗, Param Size=7B2025.11 | 86.2 | |
| ThymeTool=Code, Param Size=7B2025.11 | 86.1 | |
| TunaSize=7B2026.04 | 85.8 | |
| Tuna-RSize=7B2026.04 | 85.6 | |
| Tuna-2Size=7B2026.04 | 85.6 | |
| Qwen2.5-VL-7BModel Category=Specialist VLMs2026.04 | 84.1 | |
| Qwen2.5-VLSize=7B2026.04 | 83 | |
| X-OmniSize=7B2026.04 | 81.5 | |
| OneCatSize=9B2026.04 | 81.2 | |
| LLaVA-OV2026.04 | 80.9 | |
| LLaDA2.0-UniModel Category=Unified Models2026.04 | 80.1 | |
| LLaVA-OVTool=✗, Param Size=7B2025.11 | 80 | |
| Pre-trainedTrainable Params=02026.03 | 79.6 | |
| BAGELSize=14B2026.04 | 78.5 | |
| LLaDA-VModel Category=Specialist VLMs2026.04 | 78.3 | |
| LoRA-onlyTrainable Params=7.6M2026.03 | 78 | |
| AttnRes+LoRATrainable Params=7.7M2026.03 | 77.9 | |
| IADA +LoRATrainable Params=7.7M2026.03 | 77.9 | |
| Ming-UniVisionSize=16B2026.04 | 76.7 | |
| InternVL-UModel Category=Unified Models2026.04 | 76.6 | |
| BAGELModel Category=Unified Models2026.04 | 74.3 | |
| Emu3Size=8B2026.04 | 69.4 | |
| MRoPE2025.10 | 63.56 | |
| MHRoPE2025.10 | 62.44 | |
| MRoPE-I2025.10 | 62.12 | |
| Vanilla RoPE2025.10 | 56.84 | |
| HoPE2025.10 | 55.44 | |
| VideoRoPE2025.10 | 54.88 | |
| CircleRoPE2025.10 | 53.72 | |
| Show-o2Size=7B2026.04 | 52.3 | |
| Qwen-VL-ChatSize=7B2026.04 | 49.8 | |
| JanusFlowSize=1.3B2026.04 | 42.4 | |
| MRoPE-IBackbone Model=Qwen3-VL-4B-Instruct, Positional Encoding Variant=MRoPE-I2025.10 | 41.64 | |
| CircleRoPEBackbone Model=Qwen3-VL-4B-Instruct, Positional Encoding Variant=CircleRoPE2025.10 | 40.68 | |
| MHRoPEBackbone Model=Qwen3-VL-4B-Instruct, Positional Encoding Variant=MHRoPE2025.10 | 39.76 | |
| Vanilla RoPEBackbone Model=Qwen3-VL-4B-Instruct, Positional Encoding Variant=Vanilla RoPE2025.10 | 39.64 | |
| HoPEBackbone Model=Qwen3-VL-4B-Instruct, Positional Encoding Variant=HoPE2025.10 | 39.52 | |
| VideoRoPEBackbone Model=Qwen3-VL-4B-Instruct, Positional Encoding Variant=VideoRoPE2025.10 | 39.37 | |
| MRoPEBackbone Model=Qwen3-VL-4B-Instruct, Positional Encoding Variant=MRoPE2025.10 | 38.8 | |
| HarmonSize=1.5B2026.04 | 29.8 | |
| Janus-ProSize=7B2026.04 | 25.8 | |
| LLaVA-1.5Size=7B2026.04 | 17.8 | |
| VILA-USize=7B2026.04 | 11.4 | |
| Lumina-DiMOOModel Category=Unified Models2026.04 | 8.3 |