Health AI Benchmarks

Health Optimization Bench: incretin evidence synthesis

Can a model keep up with the evidence on GLP-1 and other incretin drugs?

More about this benchmark

Hard, freshness-dependent questions in preventive and optimization medicine, written against primary sources and graded blind by a panel.

Frontier models on hard, freshness-dependent questions in preventive and optimization medicine. Every task is written against a primary source, audited by model families that did not author it, and scored blind by a panel of independent families. The v1 release set covers incretin therapeutics evidence.

Released
Aug 2026
Snapshot
10 Sep 2026
Average panel agreement
87%

Reading these numbers

Cross-family authoring with blind three-family panel grading; the authoring family never grades its own task, and 95 percent bootstrap confidence intervals accompany every score on the site.

Scores come from runs by the benchmark's maintainers or independent researchers.

Results
18 from 17 models
Latest
Sep 2026
Published by
Health Optimization Bench
Index weight
0.226, supporting, near its ceiling
Sources
1

Counts toward the Health Evals Index as a supporting benchmark, at weight 0.226: prevention and optimization questions, one area of personal health. It is near its ceiling, with a leader at 84.9 out of 100, so it counts for less.

Where every published Health Optimization Bench: incretin evidence synthesis score sits

18 scores, one column each, highest first, on the benchmark's own scale. Higher is better.

Each column is the model's lab colour. The table below says who ran each score.

Every published score, ranked within this benchmark only

18 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 Fable 5.1Anthropic · declined 1 of 89 tasks84.9Benchmark teamSep 2026[1]
mean 0.849, ci 0.801-0.893, n 89
In the source
84.9
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Declined
declined 1 of 89 tasks
Interval
80.1 to 89.3
Document
[1] Health Optimization Bench official results
2Claude Fable 5Anthropic83.8Benchmark teamSep 2026[1]
mean 0.838, ci 0.791-0.881, n 89
In the source
83.8
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
79.1 to 88.1
Document
[1] Health Optimization Bench official results
3GPT-6 Astra (max)OpenAI · max reasoning effort82.2Benchmark teamSep 2026[1]
mean 0.822, ci 0.777-0.864, n 89
In the source
82.2
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
How it was run
max reasoning effort
Reported by
official leaderboard
Interval
77.7 to 86.4
Document
[1] Health Optimization Bench official results
4Grok 4.6SpaceX AI81.3Benchmark teamSep 2026[1]
mean 0.813, ci 0.770-0.853, n 89
In the source
81.3
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
77 to 85.3
Document
[1] Health Optimization Bench official results
5Claude Opus 5Anthropic78.3Benchmark teamSep 2026[1]
mean 0.783, ci 0.729-0.834, n 89
In the source
78.3
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
72.9 to 83.4
Document
[1] Health Optimization Bench official results
6GPT-5.6 Sol (max)OpenAI · max reasoning effort77.9Benchmark teamSep 2026[1]
mean 0.779, ci 0.733-0.822, n 89
In the source
77.9
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
How it was run
max reasoning effort
Reported by
official leaderboard
Interval
73.3 to 82.2
Document
[1] Health Optimization Bench official results
7Kimi K3Moonshot AI77.8Benchmark teamSep 2026[1]
mean 0.778, ci 0.730-0.824, n 89
In the source
77.8
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
73 to 82.4
Document
[1] Health Optimization Bench official results
8GPT-5.6 Sol (high)OpenAI · high reasoning effort77.7Benchmark teamSep 2026[1]
mean 0.777, ci 0.731-0.820, n 89
In the source
77.7
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
How it was run
high reasoning effort
Reported by
official leaderboard
Interval
73.1 to 82
Document
[1] Health Optimization Bench official results
9Muse SparkMeta68.4Benchmark teamSep 2026[1]
mean 0.684, ci 0.628-0.739, n 89
In the source
68.4
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
62.8 to 73.9
Document
[1] Health Optimization Bench official results
10Gemini 3.8 FlashGoogle67.8Benchmark teamSep 2026[1]
mean 0.678, ci 0.617-0.736, n 89
In the source
67.8
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
61.7 to 73.6
Document
[1] Health Optimization Bench official results
11Gemini 3.6Google57.9Benchmark teamSep 2026[1]
mean 0.579, ci 0.512-0.643, n 89
In the source
57.9
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
51.2 to 64.3
Document
[1] Health Optimization Bench official results
12InklingThinking Machines53.9Benchmark teamSep 2026[1]
mean 0.539, ci 0.477-0.601, n 89
In the source
53.9
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
47.7 to 60.1
Document
[1] Health Optimization Bench official results
13Claude Sonnet 5Anthropic50.1Benchmark teamSep 2026[1]
mean 0.501, ci 0.443-0.561, n 89
In the source
50.1
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
44.3 to 56.1
Document
[1] Health Optimization Bench official results
14MiniMax M3MiniMax35.5Benchmark teamSep 2026[1]
mean 0.355, ci 0.296-0.416, n 89
In the source
35.5
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
29.6 to 41.6
Document
[1] Health Optimization Bench official results
15MAI ThinkingMicrosoft AI33.0Benchmark teamSep 2026[1]
mean 0.330, ci 0.268-0.393, n 89
In the source
33.0
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
26.8 to 39.3
Document
[1] Health Optimization Bench official results
16GLM 5.2Zhipu29.2Benchmark teamSep 2026[1]
mean 0.292, ci 0.239-0.349, n 89
In the source
29.2
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
23.9 to 34.9
Document
[1] Health Optimization Bench official results
17Mistral Medium 3.5Mistral14.4Benchmark teamSep 2026[1]
mean 0.144, ci 0.106-0.185, n 89
In the source
14.4
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
10.6 to 18.5
Document
[1] Health Optimization Bench official results
18Nemotron 3.5 LightningNVIDIA7.8Benchmark teamSep 2026[1]
mean 0.078, ci 0.047-0.113, n 89
In the source
7.8
Where to find it
tasksets/glp1-ev/analysis.json, snapshot 2026-09-10
Reported by
official leaderboard
Interval
4.7 to 11.3
Document
[1] Health Optimization Bench official results
Sources

The 1 document behind these scores

  1. 1.Health Optimization Bench official resultsofficial leaderboard, Health Optimization Bench. Quotes from this document