Search on MM Search (Pass@1)
46.1Pass@1Base
Evaluation Results
| Method | Links | |
|---|---|---|
| BaseBackbone=Qwen3-VL-Thinking, Model Size=32B2026.05 | 46.1 | |
| GRPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.05 | 45.1 | |
| SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.05 | 45 | |
| SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.05 | 44 | |
| SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.05 | 43.3 | |
| BaseBackbone=Qwen3-VL-Thinking, Model Size=8B2026.05 | 42.7 | |
| SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.05 | 42.3 | |
| SFTBackbone=Qwen3-VL-Thinking, Model Size=4B2026.05 | 41.5 | |
| GRPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.05 | 41.4 | |
| SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.05 | 40.6 | |
| SFTBackbone=Qwen3-VL-Thinking, Model Size=8B2026.05 | 40.6 | |
| SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.05 | 40.3 | |
| BaseBackbone=Qwen3-VL-Thinking, Model Size=4B2026.05 | 37.8 | |
| SFTBackbone=Qwen3-VL-Thinking, Model Size=2B2026.05 | 35.9 | |
| BaseBackbone=Qwen3-VL-Thinking, Model Size=2B2026.05 | 27.7 | |
| GRPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.05 | 23.8 |