Multimodal Multi-choice on MMStar (Accuracy)
65.1AccuracyClaude3.7-Sonnet
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude3.7-Sonnet2025.12 | 65.1 | |
| SkiLa2025.12 | 64.8 | |
| SkiLa-V2025.12 | 64.5 | |
| LLaVA-OneVision 7BModel Size=7B2025.12 | 61.9 | |
| GPT-4o2025.12 | 61.6 | |
| Direct SFT2025.12 | 61.4 | |
| LVR 7B*Model Size=7B, Tested by authors=true2025.12 | 61.3 | |
| Vision-R1 7B*Model Size=7B, Tested by authors=true2025.12 | 60.7 | |
| Qwen2.5-VL 7BModel Size=7B2025.12 | 60.3 | |
| Gemma3 27BModel Size=27B2025.12 | 59.6 | |
| GPT-4v2025.12 | 56 | |
| GPT-4o-mini2025.12 | 54.8 | |
| ROSS 7BModel Size=7B2025.12 | 53.9 | |
| Cambrian 13BModel Size=13B2025.12 | 47.1 | |
| Janus-Pro 7BModel Size=7B2025.12 | 46.5 |