MM General VQA on SEEDBench IMG
80.4AccuracyK2.5
Evaluation results
| Method | Links | |
|---|---|---|
| K2.5Parameter Count=≈1T2026.06 | 80.4 | |
| Qwen3-VL-32BDecoding Mode=Thinking, Model Scale=32B2026.06 | 79.4 | |
| Qwen3.5-PlusParameter Count=397B2026.06 | 79.4 | |
| Qwen3-VL-32BDecoding Mode=Instruct, Model Scale=32B2026.06 | 79 | |
| Qwen3-VL-8BDecoding Mode=Thinking, Model Scale=8B2026.06 | 78.5 | |
| Qwen3-VL-8BDecoding Mode=Instruct, Model Scale=8B2026.06 | 77.4 | |
| GPT-5.4Parameter Count=≈1.5T2026.06 | 77.3 | |
| Opus-4.6Parameter Count=≈800B2026.06 | 77.3 | |
| Yuvion VL-32BDecoding Mode=Reasoning, Model Scale=32B2026.06 | 77 | |
| Yuvion VL-32BDecoding Mode=Instruct, Model Scale=32B2026.06 | 76.9 | |
| Yuvion VL-8BDecoding Mode=Instruct, Model Scale=8B2026.06 | 76.5 | |
| Yuvion VL-8BDecoding Mode=Reasoning, Model Scale=8B2026.06 | 76.4 |