HealthBench Hard

Accessible data for HealthBench Hard leaderboard
Model	Rank	Model ID	Provider	Score	Unit
Muse Spark	#1	muse-spark	Meta	42.8	percent
GPT-5.4	#2	gpt-5-4	OpenAI	40.1	percent
GPT-5.6 Sol	#3	gpt-5-6-sol	OpenAI	33.1	percent
GPT-5.6 Terra	#4	gpt-5-6-terra	OpenAI	32.7	percent
GPT-5.6 Luna	#5	gpt-5-6-luna	OpenAI	32	percent
Grok 4.20	#6	grok-4-20	xAI	20.3	percent
Claude Opus 4.6	#7	claude-opus-4-6	Anthropic	14.8	percent

HealthBench Hard leaderboard ranked by score
Rank	Model	Provider	License	Evidence use	Score
#1	Muse Spark muse-spark	Meta	closed	Source carrier	42.8%
#2	GPT-5.4 gpt-5.4	OpenAI	closed	Source carrier	40.1%
#3	GPT-5.6 Sol gpt-5.6-sol	OpenAI	closed	Source carrier	33.1%
#4	GPT-5.6 Terra gpt-5.6-terra	OpenAI	closed	Source carrier	32.7%
#5	GPT-5.6 Luna gpt-5.6-luna	OpenAI	closed	Source carrier	32%
#6	Grok 4.20 grok-4.20	xAI	closed	Source carrier	20.3%
#7	Claude Opus 4.6 claude-opus-4-6	Anthropic	closed	Source carrier	14.8%

Rank

Model

Provider

License

Evidence use

Score

Muse Spark

muse-spark

Every distinct dated source-published row for HealthBench Hard, including exact model configuration labels
Exact published model / configuration	Canonical product	Score	Version / system	Evidence	Source / observed
Exact BenchLM registry variant Muse Spark; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Muse Spark	42.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant GPT-5.4; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.4	40.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant GPT-5.6 Sol; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Sol	33.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant GPT-5.6 Terra; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Terra	32.7%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant GPT-5.6 Luna; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Luna	32%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant Grok 4.20; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Grok 4.20	20.3%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant Claude Opus 4.6; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Claude Opus 4.6	14.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant Muse Spark; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Muse Spark	42.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant GPT-5.4; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.4	40.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant GPT-5.6 Sol; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Sol	33.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant GPT-5.6 Terra; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Terra	32.7%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant GPT-5.6 Luna; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Luna	32%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant Grok 4.20; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Grok 4.20	20.3%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant Claude Opus 4.6; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Claude Opus 4.6	14.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant Muse Spark; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Muse Spark	42.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant GPT-5.4; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.4	40.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant GPT-5.6 Sol; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Sol	33.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant GPT-5.6 Terra; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Terra	32.7%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant GPT-5.6 Luna; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Luna	32%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant Grok 4.20; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Grok 4.20	20.3%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant Claude Opus 4.6; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Claude Opus 4.6	14.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21

What does HealthBench Hard measure?

OpenAI HealthBench hard subset evaluating model helpfulness and safety on challenging healthcare conversations.

Which model scores highest on HealthBench Hard?

Muse Spark by Meta currently leads with 42.8%.

How many models are evaluated on HealthBench Hard?

7 canonical model products have a qualifying score, backed by 21 distinct dated source rows.

Does this affect overall Lumina rank?

No. This benchmark is display-only and does not enter the overall Lumina composite.

Accessible data for HealthBench Hard leaderboard
Model	Rank	Model ID	Provider	Score	Unit
Muse Spark	#1	muse-spark	Meta	42.8	percent
GPT-5.4	#2	gpt-5-4	OpenAI	40.1	percent
GPT-5.6 Sol	#3	gpt-5-6-sol	OpenAI	33.1	percent
GPT-5.6 Terra	#4	gpt-5-6-terra	OpenAI	32.7	percent
GPT-5.6 Luna	#5	gpt-5-6-luna	OpenAI	32	percent
Grok 4.20	#6	grok-4-20	xAI	20.3	percent
Claude Opus 4.6	#7	claude-opus-4-6	Anthropic	14.8	percent

HealthBench Hard leaderboard ranked by score
Rank	Model	Provider	License	Evidence use	Score
#1	Muse Spark muse-spark	Meta	closed	Source carrier	42.8%
#2	GPT-5.4 gpt-5.4	OpenAI	closed	Source carrier	40.1%
#3	GPT-5.6 Sol gpt-5.6-sol	OpenAI	closed	Source carrier	33.1%
#4	GPT-5.6 Terra gpt-5.6-terra	OpenAI	closed	Source carrier	32.7%
#5	GPT-5.6 Luna gpt-5.6-luna	OpenAI	closed	Source carrier	32%
#6	Grok 4.20 grok-4.20	xAI	closed	Source carrier	20.3%
#7	Claude Opus 4.6 claude-opus-4-6	Anthropic	closed	Source carrier	14.8%

Rank

Model

Provider

License

Evidence use

Score

Muse Spark

muse-spark

Every distinct dated source-published row for HealthBench Hard, including exact model configuration labels
Exact published model / configuration	Canonical product	Score	Version / system	Evidence	Source / observed
Exact BenchLM registry variant Muse Spark; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Muse Spark	42.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant GPT-5.4; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.4	40.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant GPT-5.6 Sol; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Sol	33.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant GPT-5.6 Terra; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Terra	32.7%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant GPT-5.6 Luna; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Luna	32%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant Grok 4.20; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Grok 4.20	20.3%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant Claude Opus 4.6; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Claude Opus 4.6	14.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-08-01 ↗Observed 2026-08-01 · checked 2026-08-01
Exact BenchLM registry variant Muse Spark; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Muse Spark	42.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant GPT-5.4; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.4	40.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant GPT-5.6 Sol; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Sol	33.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant GPT-5.6 Terra; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Terra	32.7%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant GPT-5.6 Luna; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Luna	32%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant Grok 4.20; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Grok 4.20	20.3%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant Claude Opus 4.6; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Claude Opus 4.6	14.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27
Exact BenchLM registry variant Muse Spark; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Muse Spark	42.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant GPT-5.4; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.4	40.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant GPT-5.6 Sol; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Sol	33.1%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant GPT-5.6 Terra; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Terra	32.7%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant GPT-5.6 Luna; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	GPT-5.6 Luna	32%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant Grok 4.20; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Grok 4.20	20.3%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21
Exact BenchLM registry variant Claude Opus 4.6; bulk export does not retain a complete upstream harness configuration. source label without registered configuration ID	Claude Opus 4.6	14.8%	2026Source-native system	Source carrier · no additional scoring weight source-checked	BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21

What does HealthBench Hard measure?

OpenAI HealthBench hard subset evaluating model helpfulness and safety on challenging healthcare conversations.

Which model scores highest on HealthBench Hard?

Muse Spark by Meta currently leads with 42.8%.

How many models are evaluated on HealthBench Hard?

7 canonical model products have a qualifying score, backed by 21 distinct dated source rows.

Does this affect overall Lumina rank?

No. This benchmark is display-only and does not enter the overall Lumina composite.

Navigation

HealthBench Hard

Leaderboard · 7 models

Muse Spark

GPT-5.4

GPT-5.6 Sol