List your company

Reasoning on V* (Pass@4)

97.9Pass@4SFT + GRPO
79.90884.57989.2593.921May 27, 2026

Evaluation results

MethodLinks
SFT + GRPOModel Size=4B, Backbone=Qwen3-VL-Thinking2026.0597.9
AXPOModel Size=4B, Backbone=Qwen3-VL-Thinking2026.0597.9
BaseModel Size=32B, Backbone=Qwen3-VL-Thinking2026.0597.4
SFT + GRPOModel Size=8B, Backbone=Qwen3-VL-Thinking2026.0596.9
BaseModel Size=4B, Backbone=Qwen3-VL-Thinking2026.0595.8
SFTModel Size=4B, Backbone=Qwen3-VL-Thinking2026.0595.3
AXPOModel Size=8B, Backbone=Qwen3-VL-Thinking2026.0595.3
SFT + GRPOModel Size=2B, Backbone=Qwen3-VL-Thinking2026.0594.2
AXPOModel Size=2B, Backbone=Qwen3-VL-Thinking2026.0594.2
GRPOModel Size=8B, Backbone=Qwen3-VL-Thinking2026.0594.2
SFTModel Size=8B, Backbone=Qwen3-VL-Thinking2026.0593.7
GRPOModel Size=4B, Backbone=Qwen3-VL-Thinking2026.0592.7
BaseModel Size=8B, Backbone=Qwen3-VL-Thinking2026.0592.7
SFTModel Size=2B, Backbone=Qwen3-VL-Thinking2026.0591.1
BaseModel Size=2B, Backbone=Qwen3-VL-Thinking2026.0581.7
GRPOModel Size=2B, Backbone=Qwen3-VL-Thinking2026.0580.6