Multimodal Perception on VStar
92.67AccuracyZwZ-4B
Evaluation Results
| Method | Links | |
|---|---|---|
| ZwZ-4Bnumber of parameters=4B2026.02 | 92.67 | |
| ZwZ-8Bnumber of parameters=8B2026.02 | 91.1 | |
| ZwZModel Size=8B2026.02 | 91.1 | |
| ZwZModel Size=7B2026.02 | 88.5 | |
| ZwZ-7Bnumber of parameters=7B2026.02 | 88.48 | |
| Qwen3-VL-235B-A22Bnumber of parameters=235B2026.02 | 87.96 | |
| Gemini-3-Flash2026.02 | 86.39 | |
| Qwen3-VL-8Bnumber of parameters=8B2026.02 | 86.39 | |
| Kimi-K2.5 (1T)number of parameters=1T2026.02 | 85.86 | |
| GLM-4.5V (108B)number of parameters=108B2026.02 | 83.25 | |
| DiGModel Size=8B2026.02 | 81.2 | |
| MiMo-VL-7B-RLnumber of parameters=7B2026.02 | 81.15 | |
| AGILEModel Size=7B2026.02 | 80.6 | |
| Image-JigsawModel Size=7B2026.02 | 80.6 | |
| Qwen3-VL-4Bnumber of parameters=4B2026.02 | 80.1 | |
| Qwen2.5-VL-7Bnumber of parameters=7B2026.02 | 78.53 | |
| LLaVA-OVScale=1.5-8B2026.06 | 73.3 | |
| Qwen2.5-VLScale=7B2026.06 | 71.73 | |
| GPT-5.12026.02 | 70.16 | |
| MiniCPM-V-4.5 (9B)number of parameters=9B2026.02 | 70.16 | |
| MOSS-Video-PreviewSFT Mode=offline2026.06 | 62.83 | |
| MOSS-Video-PreviewSFT Mode=real-time2026.06 | 49.74 |