health Bench Hard

5 models evaluated

#	Model	Provider	Type	Score
1	Muse Spark	Meta	Closed	42.8%
2	GPT-5.4	OpenAI	Closed	40.1%
3	Gemini 3.1 Pro	Google	Closed	20.6%
4	Grok 4.20	xAI	Closed	20.3%
5	Claude Opus 4.6	Anthropic	Closed	14.8%