Benchmark
HealthBench Professional
Scores are percentages. Higher is better.
Benchmark sourceScores from BenchBoard docs, Clinical Benchmarks data docs, LLM Stats data.
- Models and systems measured
- 45
- Scores recorded
- 113
- Sources
- 3
- IQ weight
- 2.33
Score distribution
- 58.2%
- Mean
- 59.8%
- Median
- 8.5%
- Std dev
- 46.2%
- 10th percentile
- 66.3%
- 90th percentile
- 32.2%
- Min
This benchmark has a weight of 2.33 for chat models, measured over 43 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model or system | Score |
|---|---|---|
| 51 | GPT-5.6 SolOpus 5 System Card capability summary · adaptive thinking · max effort unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 60.5% |
| 52 | Muse Wonder· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 60.5% |
| 53 | Muse Wonderclinical-result-2405· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 60.5% |
| 54 | Muse Wonder 1.0· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 60.5% |
| 55 | Claude Opus 5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 59.8% |
| 56 | Claude Opus 5clinical-result-206· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 59.8% |
| 57 | Claude Opus 5Opus 5 System Card capability summary · adaptive thinking · max effort unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 59.8% |
| 58 | Claude Opus 5Opus 5.5 System Card · adaptive thinking · max effort unless stated · length-adjusted · same Opus 5.5 card setup· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 59.8% |
| 59 | Claude Opus 5System Card capability summary · adaptive thinking · max effort · default sampling · 5 trials · context ≤1M unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 59.8% |
| 60 | Muse Spark 1.1clinical-result-505· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 59.3% |
| 61 | Muse Spark 1.1Meta Model API · xhigh · 525 items · length-normalized · GPT-5.4 low grader· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 59.3% |
| 62 | Claude Fable 5.1length-adjusted· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 58.1% |
| 63 | Claude Sonnet 5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 57.8% |
| 64 | Claude Sonnet 5clinical-result-207· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 57.8% |
| 65 | Claude Sonnet 5length-adjusted · Opus 4.8 grader · no tools · 5 trials· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 57.8% |
| 66 | GPT-5.6 Terra· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 57.7% |
| 67 | GPT-5.6 Terraclinical-result-208· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 57.7% |
| 68 | GPT-5.6 TerraOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 57.7% |
| 69 | Claude Opus 4.8clinical-result-1871· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 57.4% |
| 70 | Claude Opus 4.8Opus 5 System Card capability summary · adaptive thinking · max effort unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 57.4% |
| 71 | Claude Opus 4.8Fable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 56.9% |
| 72 | Grok 4.7· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 56.7% |
| 73 | Grok 4.7clinical-result-1804· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 56.7% |
| 74 | Grok 4.7length-adjusted score · rejected requests scored zero · xhigh effort· Reported by SpaceXAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 56.7% |
| 75 | Claude Opus 5length-adjusted· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 56.4% |
| 76 | Claude Opus 4.8· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 55.8% |
| 77 | Claude Opus 4.8clinical-result-209· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 55.8% |
| 78 | Claude Opus 4.8max · 525 items · length-normalized · GPT-5.4 low grader· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 55.8% |
| 79 | GPT-5.6 Luna· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 55.7% |
| 80 | GPT-5.6 Lunaclinical-result-210· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 55.7% |
| 81 | GPT-5.6 LunaOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 55.7% |
| 82 | Gemini 4 Argonclinical-result-2385· Clinical Benchmarks data docs | 55.1% |
| 83 | Ling 3.0 Flash SanteLing-3.0-flash-Sante official evaluation image · highest reasoning tier · temperature 1.0 / top_p 0.95 unless otherwise stated · exact API snapshot and run count not published · professional healthcare score· Reported by Ant inclusionAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 54.3% |
| 84 | Muse Sparkclinical-result-506· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 54.1% |
| 85 | GPT-5.6 Solclinical-result-492· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 54% |
| 86 | Kimi K3clinical-result-2386· Clinical Benchmarks data docs | 53.2% |
| 87 | Claude Opus 4.8OpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 53% |
| 88 | GLM 5.3clinical-result-2387· Clinical Benchmarks data docs | 52.8% |
| 89 | Gemini 3.8 Flashlength-adjusted· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 52.1% |
| 90 | Claude Opus 4.7clinical-result-504· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 51.9% |
| 91 | GPT 5.5clinical-result-489· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 51.8% |
| 92 | GPT 5.5Fable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 51.8% |
| 93 | GPT 5.5length-adjusted · no tools · GPT-5.5 system-card implementation· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 51.8% |
| 94 | GPT 5.5xhigh · 525 items · length-normalized · GPT-5.4 low grader· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 51.8% |
| 95 | Gemini 3.8 Flashclinical-result-2388· Clinical Benchmarks data docs | 51.7% |
| 96 | DeepSeek V4.1 Flashclinical-result-2389· Clinical Benchmarks data docs | 50.7% |
| 97 | GPT 5.5OpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 49.5% |
| 98 | Grok 4.6clinical-result-1805· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 48.5% |
| 99 | GPT 5.4clinical-result-488· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 48.1% |
| 100 | GPT 5clinical-result-485· Clinical Benchmarks data docs· Reported by the model's developer. The source relaying it did not verify it. claim | 46.2% |
Showing 51–100 of 111 results.
Not ranked
The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.
| Model or system | Score |
|---|---|
| Grok 4.6length-adjusted score · rejected requests scored zero · xhigh effort· Reported by SpaceXAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 48.5% |
| Muse Spark (version unspecified)Meta Model API · highest effort · version unspecified · 525 items · length-normalized · GPT-5.4 low grader· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 54.1% |
Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Clinical Benchmarks (clinicalbenchmarks.ai), CC BY 4.0; fractions converted to percentages; source documents retain their own terms (licence) · Data by LLM Stats