Chart Question Answering on ChartQA (Score)
94.4ScoreMiMoVL 7B-RL
Evaluation results
| Method | Links | |
|---|---|---|
| MiMoVL 7B-RLRL=MiMoVL, Initial Model=7B-SFT2026.04 | 94.4 | |
| Vero Q3I-8BRL=Qwen3VL, Initial Model=8B Inst2026.04 | 91.6 | |
| Vero Q3T-8BRL=Qwen3VL, Initial Model=8B Think2026.04 | 90.8 | |
| Vero Q25-7BRL=Qwen25VL, Initial Model=7B Inst2026.04 | 90.6 | |
| Vero Mi-7BRL=MiMoVL, Initial Model=7B-SFT2026.04 | 90.4 | |
| Q3VL 8B-InsRL=N/A, Initial Model=N/A2026.04 | 89.6 | |
| Q3VL 8B-ThkRL=N/A, Initial Model=N/A2026.04 | 88.6 | |
| LLaVA OV1.5-RLRL=LLaVA, Initial Model=OV1.5-Ins2026.04 | 87.4 | |
| Q25VL 7B-InsRL=N/A, Initial Model=N/A2026.04 | 87.3 | |
| Qwen2.5-VL-3BModel Category=Autoregressive2026.04 | 84 | |
| DatologyScale=4B2026.05 | 84 | |
| InternVL3.5Scale=4B2026.05 | 84 | |
| Fast-dVLMModel Category=Diffusion, Decoding Strategy=spec.2026.04 | 83.1 | |
| Qwen3.5-VLScale=2B2026.05 | 83.1 | |
| Fast-dVLMModel Category=Diffusion, Decoding Strategy=MDM2026.04 | 82.8 | |
| Qwen3-VLScale=4B2026.05 | 82.8 | |
| Qwen3.5-VLScale=4B2026.05 | 82.3 | |
| DatologyScale=2B2026.05 | 81.7 | |
| MAmmoTH-VLScale=4B2026.05 | 81.4 | |
| GPT-5 NanoRL=N/A, Initial Model=N/A2026.04 | 80.1 | |
| MAmmoTH-VLScale=2B2026.05 | 78.9 | |
| DatologyScale=1B2026.05 | 78.7 | |
| InternVL3Scale=2B2026.05 | 78.4 | |
| LLaDA-VModel Category=Diffusion2026.04 | 78.3 | |
| Intern-VL-2.5-4BModel Category=Autoregressive2026.04 | 77.8 | |
| VanillaBackbone=InternVL-2.5-8B, Token Reduction Rate=100%2026.04 | 77.4 | |
| Qwen3-VLScale=2B2026.05 | 76.8 | |
| Mo2-O 7BRL=SFT Only, Initial Model=N/A2026.04 | 75.2 | |
| InternVL3.5Scale=2B2026.05 | 74.8 | |
| MAmmoTH-VLScale=1B2026.05 | 72.4 | |
| DASHBackbone=InternVL-2.5-8B, Token Reduction Rate=25%2026.04 | 65.34 | |
| FastVBackbone=InternVL-2.5-8B, Token Reduction Rate=25%2026.04 | 63.68 | |
| DimpleModel Category=Diffusion2026.04 | 63.3 | |
| MiniCPM-V-2 (3B)Model Category=Autoregressive2026.04 | 59.2 | |
| LaViDaModel Category=Diffusion2026.04 | 59 | |
| DivPruneBackbone=InternVL-2.5-8B, Token Reduction Rate=25%2026.04 | 58.84 | |
| VILA-1.5-3BModel Category=Autoregressive2026.04 | 53 | |
| DARTBackbone=InternVL-2.5-8B, Token Reduction Rate=25%2026.04 | 49.44 | |
| EOSRetain (visual tokens)=322026.06 | 15.36 | |
| SparseVLM-v2Retain (visual tokens)=322026.06 | 14.87 | |
| SparseVLM-v1Retain (visual tokens)=322026.06 | 14.04 | |
| ToMeRetain (visual tokens)=322026.06 | 13.05 | |
| EOSRetain (visual tokens)=162026.06 | 12.55 | |
| SparseVLM-v2Retain (visual tokens)=162026.06 | 12.04 | |
| SparseVLM-v1Retain (visual tokens)=162026.06 | 11.4 | |
| ToMeRetain (visual tokens)=162026.06 | 10.71 | |
| PoolRetain (visual tokens)=322026.06 | 9.4 | |
| PoolRetain (visual tokens)=162026.06 | 7.55 |