Artificial Analysis Healthcare & Medical Index
How do general-purpose models rank on an index weighted toward health work?
More about this benchmark
A weighted composite of health knowledge, agentic knowledge work, long-context reasoning, and non-hallucination evaluations.
Weighted composite for healthcare and medical work: Medical & Health Knowledge 30%, Agentic Knowledge Work 25%, Long-Context Reasoning 15%, Non-Hallucination 10%, Reasoning 10%, Agentic Customer Interaction 10%, drawn from AA-Omniscience Health Accuracy, GDPval-AA v2, MLCR-AA, AA-Omniscience Non-Hallucination, Humanity's Last Exam and tau3-Banking.
- Models on the official board
- 23
- Evaluations inside the index
- AA-Omniscience Health Accuracy, GDPval-AA v2, MLCR-AA, AA-Omniscience Non-Hallucination, Humanity's Last Exam, tau3-Banking
- Official board updated
- 7 Sep 2026
- Rounding on the official board
- whole numbers (raw floats kept in extra.raw)
Reading these numbers
A weighted composite of general-purpose benchmarks tilted toward health-relevant slices rather than purpose-built clinical tasks, which is why this index files it as a capability index. Artificial Analysis maintains the index, so its page is the official board. The composition changed since this row was first written: the page now says five evaluations and adds a 15% Long-Context Reasoning capability (MLCR-AA), with Non-Hallucination and Reasoning cut to 10% each and Medical & Health Knowledge to 30%. The board is JavaScript rendered and prints scores as whole numbers.
Scores come from runs by the benchmark's maintainers or independent researchers.
- Results
- 29 from 29 models
- Latest
- Sep 2026
- Published by
- Artificial Analysis
- Released
- Aug 2026
- Sources
- 1
Where every published Artificial Analysis Healthcare & Medical Index score sits
29 scores, one column each, highest first, on the benchmark's own scale. Higher is better.
- Claude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)
- Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)
- Claude Opus 5 (Adaptive Reasoning, Max Effort)
- Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)
- GPT-6 Astra (max)
- Muse Spark 1.3 (max)
- GLM-5.3 (max)
- Grok 4.7 (xhigh)
- GLM 5.3 Flash
- GPT-5.6 Sol (max)
- Kimi K3 (max)
- Grok 4.6 (high)
- GPT-6 Sol (max)
- Gemini 3.8 Flash (high)
- MiMo-V2.6-Pro
- DeepSeek V4.1 Flash (Reasoning, Max Effort)
- Qwen3.8 Max (0902)
- Step 5 Preview
- GPT-5.6 Terra (max)
- GPT-6 Luna (max)
- GPT-5.6 Luna (max)
- Qwen3.8 27B (xhigh)
- MiniMax-M3
- Inkling (xhigh)
- Gemini 3.5 Flash-Lite
Each column is the model's lab colour. The table below says who ran each score.
Every published score, ranked within this benchmark only
29 scores. Higher is better. Where the source gave a range, the whisker shows it. Open any row to read the sentence the score came from.
1Claude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)Anthropic61Benchmark teamSep 2026[1]
Claude Opus 5.5 (max with fallback) | 61
2Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)Anthropic58Benchmark teamSep 2026[1]
Claude Fable 5.1 (max with fallback) | 58
3Claude Opus 5 (Adaptive Reasoning, Max Effort)Anthropic · all underlying benchmarks…53Benchmark teamSep 2026[1]
Claude Opus 5 (max) | 53
4Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)Anthropic52Benchmark teamSep 2026[1]
Based on the Artificial Analysis Healthcare & Medical Index, the top-performing AI models for healthcare and medical work are currently Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback) (56), Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) (52), and Claude Opus 5 (Adaptive Reasoning, Xhigh Effort) (52). Rankings are updated as new models are released.
4GPT-6 Astra (max)OpenAI52Benchmark teamSep 2026[1]
GPT-6 Astra (max) | 52
6Muse Spark 1.3 (max)Meta50Benchmark teamSep 2026[1]
Muse Spark 1.3 (max) | 50
7GLM-5.3 (max)Zhipu47Benchmark teamSep 2026[1]
GLM-5.3 (max) | 47
7Grok 4.7 (xhigh)SpaceX AI47Benchmark teamSep 2026[1]
Grok 4.7 (xhigh) | 47
9GLM 5.3 FlashZhipu45Benchmark teamSep 2026[1]
GLM-5.3-Flash | 45
9GPT-5.6 Sol (max)OpenAI45Benchmark teamSep 2026[1]
GPT-5.6 Sol (max) | 45
9Kimi K3 (max)Moonshot AI45Benchmark teamSep 2026[1]
Kimi K3 (max) | 45
12Grok 4.6 (high)SpaceX AI44Benchmark teamSep 2026[1]
{"label":"Grok 4.6 (high)","score":43.92324392614704,"detailsUrl":"/models/grok-4-6"}
13GPT-6 Sol (max)OpenAI43Benchmark teamSep 2026[1]
GPT-6 Sol (max) | 43
14Gemini 3.8 Flash (high)Google42Benchmark teamSep 2026[1]
Gemini 3.8 Flash (high) | 42
14MiMo-V2.6-ProXiaomi42Benchmark teamSep 2026[1]
MiMo-V2.6-Pro | 42
16DeepSeek V4.1 Flash (Reasoning, Max Effort)DeepSeek41Benchmark teamSep 2026[1]
DeepSeek V4.1 Flash (max) | 41
16Qwen3.8 Max (0902)Alibaba41Benchmark teamSep 2026[1]
Qwen3.8 Max (0902) | 41
16Step 5 PreviewStepFun41Benchmark teamSep 2026[1]
Step 5 Preview | 41
19GPT-5.6 Terra (max)OpenAI39Benchmark teamSep 2026[1]
{"label":"GPT-5.6 Terra (max)","score":38.66673219960154,"detailsUrl":"/models/gpt-5-6-terra"}
20GPT-6 Luna (max)OpenAI37Benchmark teamSep 2026[1]
GPT-6 Luna (max) | 37
21GPT-5.6 Luna (max)OpenAI36Benchmark teamSep 2026[1]
GPT-5.6 Luna (max) | 36
22Qwen3.8 27B (xhigh)Alibaba34Benchmark teamSep 2026[1]
Qwen3.8 27B (xhigh) | 34
23MiniMax-M3MiniMax30Benchmark teamSep 2026[1]
MiniMax-M3 | 30
24Inkling (xhigh)Thinking Machines25Benchmark teamSep 2026[1]
Inkling | 25
25Gemini 3.5 Flash-LiteGoogle23Benchmark teamSep 2026[1]
Gemini 3.5 Flash-Lite | 23
25Nemotron 3 Ultra 550B A55B (Reasoning)NVIDIA23Benchmark teamSep 2026[1]
Nemotron 3 Ultra | 23
27Muse Glimmer (high)Meta18Benchmark teamSep 2026[1]
Muse Glimmer (high) | 18
28gpt-oss-120b (high)OpenAI14Benchmark teamSep 2026[1]
\"slug\":\"gpt-oss-120b\",\"name\":\"gpt-oss-120b (high)\" ... \"headlineValue\":14.423032488856672
28Mistral Medium 3.5Mistral14Benchmark teamSep 2026[1]
Mistral Medium 3.5 | 14
The 1 document behind these scores
- 1.Best AI for Healthcare & Medical: LLM Leaderboardofficial leaderboard, Artificial Analysis. Quotes from this document