Share your thoughts, 1 month free Claude Pro on usSee more

Real World Tasks on HealthBench Hard

38.6Score

GPT-5-mini High

Updated 23d ago

Evaluation Results

Method	Links
GPT-5-mini High 2026.06		38.6
GPT-5.2 High 2026.06		36.6
Seed2.0 Pro 2026.06		28.3
Gemini-3-Flash High 2026.06		21.5
Seed2.0 Lite 2026.06		20
Gemini-3-pro High 2026.06		15
Claude-Opus-4.5 2026.06		11
Claude-Sonnet-4.5 2026.06		10.9