Multimodal Perception on CV-Bench
89.57AccuracyGemini-3-Flash
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-3-Flash2026.02 | 89.57 | |
| Kimi-K2.5 (1T)number of parameters=1T2026.02 | 89.18 | |
| ZwZ-4Bnumber of parameters=4B2026.02 | 87.9 | |
| GLM-4.5V (108B)number of parameters=108B2026.02 | 87.59 | |
| ZwZ-8Bnumber of parameters=8B2026.02 | 87.4 | |
| Qwen3-VL-235B-A22Bnumber of parameters=235B2026.02 | 86.72 | |
| Qwen3-VL-8Bnumber of parameters=8B2026.02 | 85.44 | |
| Qwen3-VL-4Bnumber of parameters=4B2026.02 | 84.95 | |
| MiMo-VL-7B-RLnumber of parameters=7B2026.02 | 84.31 | |
| GPT-5.12026.02 | 84.22 | |
| LLaVA-OVScale=1.5-8B2026.06 | 80.82 | |
| MiniCPM-V-4.5 (9B)number of parameters=9B2026.02 | 80.25 | |
| Qwen2.5-VLScale=7B2026.06 | 80.14 | |
| ZwZ-7Bnumber of parameters=7B2026.02 | 79.83 | |
| Qwen2.5-VL-7Bnumber of parameters=7B2026.02 | 75.34 | |
| MOSS-Video-PreviewSFT Mode=real-time2026.06 | 73.05 | |
| MOSS-Video-PreviewSFT Mode=offline2026.06 | 70.17 |