List your company

Visual Reasoning on VisualProbe Hard

0.434AccuracyDeepconf
0.099120.186060.2730.35994Feb 13, 2026

Evaluation results

MethodLinks
DeepconfBackbone=Qwen3-VL Instruct2026.020.43414.7
SCBackbone=Qwen3-VL Instruct2026.020.425—
ASCBackbone=Qwen3-VL Instruct2026.020.42522.5
ESCBackbone=Qwen3-VL Instruct2026.020.4256.7
RTWIBackbone=Qwen3-VL Instruct2026.020.42516.2
CISCBackbone=Qwen3-VL Instruct2026.020.3969.2
Self-Cer.Backbone=Qwen3-VL Instruct2026.020.37712
DeepEyesBackbone=DeepEyes-7B2026.020.351—
BaseBackbone=Qwen3-VL Instruct2026.020.349—
SCBackbone=Qwen3-VL Thinking2026.020.302—
ASCBackbone=Qwen3-VL Thinking2026.020.30229.7
ESCBackbone=Qwen3-VL Thinking2026.020.3028.6
RTWIBackbone=Qwen3-VL Thinking2026.020.30244.6
DeepconfBackbone=Qwen3-VL Thinking2026.020.29240.3
BaseBackbone=Qwen3-VL Thinking2026.020.283—
CISCBackbone=Qwen3-VL Thinking2026.020.27415
Self-Cer.Backbone=Qwen3-VL Thinking2026.020.27414.5
GPT-4oBackbone=GPT-4o2026.020.112—