Multimodal Perception on MMStar
83.6AccuracyGemini-3-Flash
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-3-Flash2026.02 | 83.6 | |
| Kimi-K2.5 (1T)number of parameters=1T2026.02 | 81.8 | |
| Qwen3-VL-235B-A22Bnumber of parameters=235B2026.02 | 76.33 | |
| GLM-4.5V (108B)number of parameters=108B2026.02 | 75.87 | |
| MiMo-VL-7B-RLnumber of parameters=7B2026.02 | 73.53 | |
| ZwZ-8Bnumber of parameters=8B2026.02 | 73.13 | |
| ZwZModel Size=8B2026.02 | 73.1 | |
| DiGModel Size=8B2026.02 | 72.7 | |
| GPT-5.12026.02 | 71.6 | |
| ZwZ-4Bnumber of parameters=4B2026.02 | 71.13 | |
| Qwen3-VL-8Bnumber of parameters=8B2026.02 | 70.93 | |
| Qwen3-VL-4Bnumber of parameters=4B2026.02 | 69.73 | |
| MiniCPM-V-4.5 (9B)number of parameters=9B2026.02 | 67.87 | |
| LLaVA-OVScale=1.5-8B2026.06 | 67.72 | |
| Qwen2.5-VLScale=7B2026.06 | 63.9 | |
| ZwZ-7Bnumber of parameters=7B2026.02 | 63.4 | |
| ZwZModel Size=7B2026.02 | 63.4 | |
| Qwen2.5-VL-7Bnumber of parameters=7B2026.02 | 61.93 | |
| Llama-3.2 11B-VisionCheckpoint=base2026.06 | 53.87 | |
| MOSS-Video-PreviewSFT Mode=real-time2026.06 | 53.11 | |
| MOSS-Video-PreviewSFT Mode=offline2026.06 | 48.99 |