List your company

Search on MM Search (Pass@1)

46.1Pass@1Base
22.90828.92934.9540.971May 27, 2026

Evaluation results

MethodLinks
BaseBackbone=Qwen3-VL-Thinking, Model Size=32B2026.0546.1
GRPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0545.1
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0545
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0544
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0543.3
BaseBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0542.7
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0542.3
SFTBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0541.5
GRPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0541.4
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0540.6
SFTBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0540.6
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0540.3
BaseBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0537.8
SFTBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0535.9
BaseBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0527.7
GRPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0523.8