List your company

Mathematical Reasoning on Math Vision (Pass@1)

64.1Pass@1Base
15.01227.75640.553.244May 27, 2026

Evaluation results

MethodLinks
BaseBackbone=Qwen3-VL-Thinking, Model Size=32B2026.0564.1
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0560.6
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0560.4
GRPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0558.9
SFTBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0558.4
GRPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0557.9
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0557.4
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0557.2
BaseBackbone=Qwen3-VL-Thinking, Model Size=32B2026.0556.5
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0556.1
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0555.3
SFTBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0555.3
BaseBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0554.9
GRPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0554.4
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0552.8
GRPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0552.5
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0551.6
SFTBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0550.9
BaseBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0550.8
SFTBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0549.1
BaseBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0548
BaseBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0547.1
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0538.9
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0535.8
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0535
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0534.8
GRPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0534.6
GRPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0533.7
SFTBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0530.2
SFTBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0528.9
BaseBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0518.6
BaseBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0516.9