Chart Understanding and Reasoning on ChartQA
89.6AccuracyInternVL3.5-38B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| InternVL3.5-38BModel Scale=Large, Framework=Baseline, Visual Highlighting=false2026.04 | 89.6 | — | |
| InternVL3.5-38B + LoTModel Scale=Large, Framework=LoT, Visual Highlighting=true2026.04 | 89 | — | |
| InternVL2.5-MPO + SFTSize=8B2026.06 | 88.3 | 71 | |
| MMS-PRMSize=8B2026.06 | 87.2 | 73 | |
| InternVL-3-8BTraining Strategy=Open-Source SFT2025.08 | 86.6 | — | |
| Insight-V++Size=7B, Base Model=Qwen2.5-VL, Self-Evolving=true2026.03 | 86.1 | — | |
| Qwen2.5-VL + Multi-Agent (RL)Size=7B, Multi-Agent (RL)=true2026.03 | 85.9 | — | |
| InternVL3.5-4BModel Scale=Small, Framework=Baseline, Visual Highlighting=false2026.04 | 85.9 | — | |
| GazeVLM (Ours)Gaze Bias=Enabled, Base Model=Qwen3-VL-4B2026.05 | 85.8 | — | |
| GazeVLM (w/o gaze bias)Gaze Bias=None, Base Model=Qwen3-VL-4B2026.05 | 85.5 | — | |
| DUELRole=Solver2026.05 | 85.5 | — | |
| InternVL3.5-4B + LoTModel Scale=Small, Framework=LoT, Visual Highlighting=true2026.04 | 85.4 | — | |
| EvoLMM2026.05 | 85.1 | — | |
| InternVL2.5-MPOSize=8B2026.06 | 85 | 69.4 | |
| Vision-Zero(CLEVR)2026.05 | 84.9 | — | |
| InternVL-2.5Size=8B2026.06 | 84.9 | 68 | |
| MM-UPT2026.05 | 84.7 | — | |
| Qwen2.5-VLSize=7B2026.03 | 84.5 | — | |
| OpenVLThinker-7B2026.05 | 84.5 | — | |
| DUELRole=Challenger2026.05 | 84.4 | — | |
| InternVL3.5-8BModel Scale=Medium, Framework=Baseline, Visual Highlighting=false2026.04 | 84.3 | — | |
| Shuffle-R1-Qwen-7BTraining Strategy=Zero RL2025.08 | 84.1 | — | |
| Qwen2.5-VL-7B + DUPLBase Model=Qwen2.5-VL-7B, RL Algorithm=DUPL2025.10 | 84 | — | |
| Vision-R1-LlamaV-CI-11BTraining Paradigm=SFT-based, Parameters=11B2026.05 | 83.9 | — | |
| Vision-R1-LlamaVSize=11B2026.06 | 83.9 | — | |
| R1-VLSize=7B2026.06 | 83.9 | — | |
| VLAA-Thinker-7B2026.05 | 83.8 | — | |
| Dynamo (Full)Model=Doubao-Seed-2.0, Evolution Strategy=Full2026.06 | 83.47 | — | |
| Qwen2.5-VL-7B2026.05 | 83.4 | — | |
| Qwen2.5-VL-7BTraining Paradigm=Vanilla Open-source, Parameters=7B2026.05 | 83.3 | — | |
| Qwen3-VL-4BTraining Paradigm=Vanilla Open-source, Parameters=4B2026.05 | 83.2 | — | |
| Vision-R1-7BTraining Strategy=Cold-Start + RL, Evaluation Toolkit=vLLM with custom scripts2025.08 | 83.1 | — | |
| Qwen2-VLSize=7B2026.03 | 83 | — | |
| Dynamo (Full)Model=GPT-5.4, Evolution Strategy=Full2026.06 | 82.76 | — | |
| Qwen3-VL-8B-InstructStrategy=Instruct2026.02 | 82.7 | — | |
| Qwen3-VL-4B + LoTModel Scale=Small, Framework=LoT, Visual Highlighting=true2026.04 | 82.3 | — | |
| IXC-2.5Size=7B2026.03 | 82.2 | — | |
| IXC-2.5Size=7B2026.06 | 82.2 | — | |
| MMR1-Math-7BTraining Strategy=Zero RL, Evaluation Toolkit=vLLM with custom scripts2025.08 | 82 | — | |
| ThinkLite-VL-7BTraining Strategy=Zero RL, Evaluation Toolkit=vLLM with custom scripts2025.08 | 82 | — | |
| Qwen2-VL-7B + LoTModel Scale=Medium, Framework=LoT, Visual Highlighting=true2026.04 | 82 | — | |
| Qwen3-VL-8BModel Scale=Medium, Framework=Baseline, Visual Highlighting=false2026.04 | 82 | — | |
| Dynamo (Skill Only)Model=Doubao-Seed-2.0, Evolution Strategy=Skill Only2026.06 | 81.98 | — | |
| Qwen2.5-VL-7B + NoisyRolloutBase Model=Qwen2.5-VL-7B, RL Algorithm=NoisyRollout2025.10 | 81.8 | — | |
| Qwen2-VL-7BModel Scale=Medium, Framework=Baseline, Visual Highlighting=false2026.04 | 81.6 | — | |
| Dynamo (Full)Model=Qwen3.5-27B, Evolution Strategy=Full2026.06 | 81.58 | — | |
| Dynamo (Tool Only)Model=Doubao-Seed-2.0, Evolution Strategy=Tool Only2026.06 | 81.56 | — | |
| Base AgentModel=Doubao-Seed-2.0, Evolution Strategy=None2026.06 | 81.52 | — | |
| Insight-VSize=7B, Base Model=Our Base Model, Iterative DPO=true2026.03 | 81.5 | — | |
| Insight-VSize=7B2026.06 | 81.5 | 65.7 | |
| Qwen2.5-VL-7B + GRPOBase Model=Qwen2.5-VL-7B, RL Algorithm=GRPO2025.10 | 81.5 | — | |
| NoisyRollout-7B-K12Training Strategy=Zero RL, Evaluation Toolkit=vLLM with custom scripts2025.08 | 81.4 | — | |
| Gemini-1.52026.02 | 81.3 | — | |
| Our Base Model + Multi-AgentSize=7B, Multi-Agent=true2026.03 | 81.2 | — | |
| Dynamo (Tool Only)Model=Qwen3.5-27B, Evolution Strategy=Tool Only2026.06 | 81.15 | — | |
| Base AgentModel=Qwen3.5-27B, Evolution Strategy=None2026.06 | 81.14 | — | |
| Claude-3-SonnetTable Header=Claude-S2026.02 | 81.1 | — | |
| Dynamo (Tool Only)Model=GPT-5.4, Evolution Strategy=Tool Only2026.06 | 81.08 | — | |
| Dynamo (Skill Only)Model=Qwen3.5-27B, Evolution Strategy=Skill Only2026.06 | 80.97 | — | |
| Dynamo (Skill Only)Model=GPT-5.4, Evolution Strategy=Skill Only2026.06 | 80.9 | — | |
| Claude-3-OpusTable Header=Claude-O2026.02 | 80.8 | — | |
| Qwen3-VL-4BModel Scale=Small, Framework=Baseline, Visual Highlighting=false2026.04 | 80.7 | — | |
| VLAA-Thinker-7BBase Model=VLAA-Thinker-7B2025.10 | 80.2 | — | |
| VLAA-Thinker-7BTraining Strategy=Cold-Start + RL, Evaluation Toolkit=vLLM with custom scripts2025.08 | 80.1 | — | |
| Dynamo (Full)Model=o4-mini, Evolution Strategy=Full2026.06 | 80.07 | — | |
| LLaVA-OneVisionSize=7B2026.06 | 80 | 64.6 | |
| Shuffle-R1-Qwen-3BTraining Strategy=Zero RL2025.08 | 79.9 | — | |
| MM-Eureka-Qwen-7BBase Model=MM-Eureka-Qwen-7B2025.10 | 79.9 | — | |
| Qwen2.5-VL-7BTraining Strategy=Open-Source SFT, Evaluation Toolkit=vLLM with custom scripts2025.08 | 79.8 | — | |
| Qwen2.5-VL-7BBase Model=Qwen2.5-VL-7B2025.10 | 79.8 | — | |
| Base AgentModel=GPT-5.4, Evolution Strategy=None2026.06 | 79.74 | — | |
| InternVL3.5-8B + LoTModel Scale=Medium, Framework=LoT, Visual Highlighting=true2026.04 | 79.7 | — | |
| PAPO-7BBase Model=PAPO-7B2025.10 | 79.6 | — | |
| Qwen2.5-VL-3B + LoTModel Scale=Small, Framework=LoT, Visual Highlighting=true2026.04 | 79.5 | — | |
| Qwen2.5-VL-7B + LoTModel Scale=Medium, Framework=LoT, Visual Highlighting=true2026.04 | 79.4 | — | |
| Qwen3-VL-8B + LoTModel Scale=Medium, Framework=LoT, Visual Highlighting=true2026.04 | 79.4 | — | |
| MiniCPM-V-2-6Size=8B2026.06 | 79.4 | 64.2 | |
| Dynamo (Tool Only)Model=o4-mini, Evolution Strategy=Tool Only2026.06 | 79.2 | — | |
| InternVL-2.5-8BTraining Strategy=Open-Source SFT2025.08 | 79.1 | — | |
| Qwen2.5-VL-3BModel Scale=Small, Framework=Baseline, Visual Highlighting=false2026.04 | 79.1 | — | |
| MM-Eureka-Qwen-7BTraining Strategy=Zero RL, Evaluation Toolkit=vLLM with custom scripts2025.08 | 79 | — | |
| VL-Rethinker-7BTraining Strategy=Zero RL, Evaluation Toolkit=vLLM with custom scripts2025.08 | 79 | — | |
| ProLaViT (Full)Reasoning Paradigm=Progressive Latent Visual Thought, Reasoning Space=Latent, Training Strategy=Distance-Weighted Diversity Loss2026.07 | 78.99 | — | |
| OpenVLThinker-7BBase Model=OpenVLThinker-7B2025.10 | 78.9 | — | |
| Qwen3.5-4B-InstTraining Paradigm=Vanilla Open-source, Parameters=4B2026.05 | 78.8 | — | |
| Dynamo (Skill Only)Model=o4-mini, Evolution Strategy=Skill Only2026.06 | 78.66 | — | |
| MM-1.5Size=7B2026.03 | 78.6 | — | |
| GPT-4V2026.02 | 78.5 | — | |
| Qwen2.5-VL-InstructReasoning Paradigm=Base Model, Reasoning Space=N/A2026.07 | 78.45 | — | |
| OpenVLThinker-7BTraining Strategy=Cold-Start + RL, Evaluation Toolkit=vLLM with custom scripts2025.08 | 78.4 | — | |
| Qwen2.5-VL-3B + DUPLBase Model=Qwen2.5-VL-3B, RL Algorithm=DUPL2025.10 | 78.4 | — | |
| Base AgentModel=o4-mini, Evolution Strategy=None2026.06 | 78.33 | — | |
| R1-Onevision-7BBase Model=R1-Onevision-7B2025.10 | 78.3 | — | |
| SAPStrategy=SAP2026.02 | 78.1 | — | |
| Dynamo (Full)Model=GPT-4o, Evolution Strategy=Full2026.06 | 77.99 | — | |
| Dynamo (Skill Only)Model=GPT-4o, Evolution Strategy=Skill Only2026.06 | 77.85 | — | |
| R1-OneVision-7BTraining Strategy=Cold-Start + RL, Evaluation Toolkit=vLLM with custom scripts2025.08 | 77.8 | — | |
| Qwen2.5-VL-3B + NoisyRolloutBase Model=Qwen2.5-VL-3B, RL Algorithm=NoisyRollout2025.10 | 77.7 | — | |
| Qwen2.5-VL-3B + GRPOBase Model=Qwen2.5-VL-3B, RL Algorithm=GRPO2025.10 | 77.6 | — | |
| Insight-V-LLaVASize=8B, Base Model=LLaVA-NeXT-LLaMA3, Iterative DPO=true2026.03 | 77.4 | — |