General Image Reasoning and Understanding on General Reasoning and Perception Aggregates
71.1Average Reasoning ScoreInsight-V++
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Insight-V++Size=7B, Base Model=Qwen2.5-VL, Self-Evolving=true2026.03 | 71.1 | 82.8 | |
| Qwen2.5-VL + Multi-Agent (RL)Size=7B, Multi-Agent (RL)=true2026.03 | 69.5 | 82.6 | |
| Qwen2.5-VLSize=7B2026.03 | 66.3 | 82.9 | |
| Insight-VSize=7B, Base Model=Our Base Model, Iterative DPO=true2026.03 | 60.1 | 78 | |
| Our Base Model + Multi-AgentSize=7B, Multi-Agent=true2026.03 | 59 | 78.4 | |
| Ovis1.5-LLaMA3Size=8B2026.03 | 57.5 | — | |
| Our Base ModelSize=7B2026.03 | 56.8 | 75 | |
| Idefics3-LLaMA3Size=8B2026.03 | 56 | — | |
| Insight-V-LLaVASize=8B, Base Model=LLaVA-NeXT-LLaMA3, Iterative DPO=true2026.03 | 54.9 | 72.3 | |
| LLaVA-NeXT-LLaMA3 + Multi-AgentSize=8B, Multi-Agent=true2026.03 | 51.8 | 70.5 | |
| LLaVA-NeXT-LLaMA3Size=8B2026.03 | 46.8 | 65.5 | |
| MiniCPM-V-2.6Size=7B2026.03 | — | 81.7 |