General VQA on MMVet
83.9ScoreBee-8B
Evaluation results
| Method | Links | |
|---|---|---|
| Bee-8BOpenness=fully open, Model Size=8B, Training Strategy=RL2025.10 | 83.9 | |
| Bee-8BOpenness=fully open, Model Size=8B, Training Strategy=SFT2025.10 | 83.3 | |
| InternVL3.5Openness=semi-open, Model Size=8B2025.10 | 83.1 | |
| Keye-VLOpenness=semi-open, Model Size=8B2025.10 | 79 | |
| InternVL3-9BModel Category=Open-Source MLLMs, Scale=9B2026.04 | 76.2 | |
| EVE (Ours-8B-iter4)Model Category=Pseudo-label-based Self-evolution Methods, Scale=8B, Iteration=42026.04 | 71.9 | |
| MM-Zero-8B-iter3Model Category=Pseudo-label-based Self-evolution Methods, Scale=8B, Iteration=32026.04 | 69.5 | |
| Jigsaw-R1-8BModel Category=Template-based Self-evolution Methods, Scale=8B2026.04 | 69.4 | |
| VisPlay-8B-iter3Model Category=Pseudo-label-based Self-evolution Methods, Scale=8B, Iteration=32026.04 | 69.2 | |
| GPT-4o-20240513Model Category=Closed-Source MLLMs2026.04 | 69.1 | |
| Qwen2.5-VLOpenness=semi-open, Model Size=7B2025.10 | 67.1 | |
| Qwen2.5-VL# Params=7B, Model Type=Understanding Only2026.05 | 67.1 | |
| InternVL2.5-MPOModel Size=8B2025.01 | 66.8 | |
| Qwen3-VL-8B-InstructModel Category=Open-Source MLLMs, Scale=8B2026.04 | 66.8 | |
| MetaQuery-XL# Params=7B, Model Type=Unified model2026.05 | 66.6 | |
| Qwen-ViPER-7BModel Category=Pseudo-label-based Self-evolution Methods, Scale=7B2026.04 | 65.8 | |
| InternVL2.5-MPOModel Size=2B2025.01 | 64 | |
| MetaQuery-L# Params=3B, Model Type=Unified model2026.05 | 63.2 | |
| InternVL2.5# Params=7B, Model Type=Understanding Only2026.05 | 62.8 | |
| BAGEL# Params=7B, Model Type=Unified model2026.05 | 62.7 | |
| Qwen2-VLModel Size=8B2025.01 | 62.6 | |
| BAGEL+Ours# Params=7B, Model Type=Unified model2026.05 | 62.1 | |
| Qwen2.5-VL# Params=3B, Model Type=Understanding Only2026.05 | 61.8 | |
| LLaVA-OneVision-72BModel Category=Open-Source MLLMs, Scale=72B2026.04 | 60.6 | |
| InternVL2.5# Params=4B, Model Type=Understanding Only2026.05 | 60.6 | |
| DeepSeekVL-2Model Size=8B2025.01 | 60.3 | |
| BLIP3-o# Params=4B, Model Type=Unified model2026.05 | 60.1 | |
| DeepSeek-VL2# Params=4.1B/28B, Model Type=Understanding Only2026.05 | 60 | |
| AutoVBackbone=Qwen2.5-VL 7B2025.06 | 59.1 | |
| SAIL-VLModel Size=8B2025.01 | 58.3 | |
| APIBackbone=Qwen2.5-VL 7B2025.06 | 58 | |
| LLaVA OneVisionOpenness=fully open, Model Size=7B2025.10 | 57.5 | |
| Llava-OV# Params=7B, Model Type=Understanding Only2026.05 | 57.5 | |
| BaseBackbone=Qwen2.5-VL 7B2025.06 | 56.6 | |
| AutoVBackbone=LLaVA-OneVision 7B2025.06 | 52.9 | |
| DeepSeekVL-2Model Size=2B2025.01 | 52.8 | |
| APIBackbone=LLaVA-OneVision 7B2025.06 | 51.5 | |
| Qwen2-VLModel Size=2B2025.01 | 51.2 | |
| AutoVBackbone=InternVL2 8B2025.06 | 50.2 | |
| Janus-Pro# Params=7B, Model Type=Unified model2026.05 | 50 | |
| FGVPBackbone=LLaVA-OneVision 7B2025.06 | 49.1 | |
| BaseBackbone=LLaVA-OneVision 7B2025.06 | 48.8 | |
| APIBackbone=InternVL2 8B2025.06 | 48.6 | |
| CircleBackbone=LLaVA-OneVision 7B2025.06 | 47.2 | |
| BaseBackbone=InternVL2 8B2025.06 | 46.5 | |
| SAIL-VLModel Size=2B2025.01 | 46.1 | |
| CircleBackbone=InternVL2 8B2025.06 | 45.8 | |
| CircleBackbone=Qwen2.5-VL 7B2025.06 | 44.5 | |
| FGVPBackbone=InternVL2 8B2025.06 | 44.3 | |
| FGVPBackbone=Qwen2.5-VL 7B2025.06 | 43.1 | |
| MolmoOpenness=fully open, Model Size=7B, Training Strategy=Distilled (-D)2025.10 | 41.5 | |
| TokenFlow-XL# Params=13B, Model Type=Unified model2026.05 | 40.7 | |
| AutoVBackbone=LLaVA-1.5 13B2025.06 | 38 | |
| APIBackbone=LLaVA-1.5 13B2025.06 | 37.6 | |
| Emu3-chat# Params=8B, Model Type=Understanding Only2026.05 | 37.2 | |
| BaseBackbone=LLaVA-1.5 13B2025.06 | 36.1 | |
| AutoVBackbone=LLaVA-1.5 7B2025.06 | 32.9 | |
| APIBackbone=LLaVA-1.5 7B2025.06 | 31.8 | |
| CircleBackbone=LLaVA-1.5 7B2025.06 | 31.3 | |
| CircleBackbone=LLaVA-1.5 13B2025.06 | 31.2 | |
| BaseBackbone=LLaVA-1.5 7B2025.06 | 30.6 | |
| FGVPBackbone=LLaVA-1.5 13B2025.06 | 29.7 | |
| FGVPBackbone=LLaVA-1.5 7B2025.06 | 29 |