List your company

Visual Reasoning on HR-Bench 4K FCP

63.3AccuracyGPT-4o
56.22858.06459.961.736Feb 13, 2026

Evaluation results

MethodLinks
GPT-4o2026.0263.3—
Thyme2026.0263—
RTWIBackbone=DeepEyes-7B2026.026053.4
SCBackbone=DeepEyes-7B2026.0258—
ASCBackbone=DeepEyes-7B2026.025848.9
ESCBackbone=DeepEyes-7B2026.025831.3
DeepconfBackbone=DeepEyes-7B2026.025848
Self-Uncer.Backbone=DeepEyes-7B2026.025840.2
CISCBackbone=DeepEyes-7B2026.0256.837.1
Base*Backbone=DeepEyes-7B2026.0256.5—