Skip to content

Benchmark

Bedside Bench

Scores are percentages. Higher is better.

Models measured
15
Scores recorded
15
Sources
1
IQ weight
1.14

Score distribution

88.4%
Mean
88.1%
Median
4.1%
Std dev
82.9%
10th percentile
94.3%
90th percentile
80.6%
Min
80.6%94.6%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.14 for chat models, measured over 15 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Claude Opus 5clinical-result-2551· Clinical Benchmarks data docs94.6%
2Muse Spark Openclinical-result-2552· Clinical Benchmarks data docs94.3%
3GPT-6 Astraclinical-result-2553· Clinical Benchmarks data docs93.4%
4Gemini 4 Argonclinical-result-2554· Clinical Benchmarks data docs92.7%
5GPT-5.6 Solclinical-result-2555· Clinical Benchmarks data docs91.1%
6Kimi K3clinical-result-2556· Clinical Benchmarks data docs89.7%
7Ember-1clinical-result-2557· Clinical Benchmarks data docs89.1%
8GPT-6 Solclinical-result-2558· Clinical Benchmarks data docs88.1%
9GPT-6 Lunaclinical-result-2559· Clinical Benchmarks data docs87.2%
10GLM 5.3clinical-result-2560· Clinical Benchmarks data docs86.9%
11Gemini 3.8 Flashclinical-result-2561· Clinical Benchmarks data docs86.5%
12DeepSeek V4.1 Flash 1017clinical-result-2562· Clinical Benchmarks data docs85.7%
13Grok 4.6clinical-result-2563· Clinical Benchmarks data docs83.3%
14DeepSeek V4.1 Flashclinical-result-2564· Clinical Benchmarks data docs82.9%
15DeepSeek V4 Proclinical-result-2565· Clinical Benchmarks data docs80.6%

Data credits: Clinical Benchmarks (clinicalbenchmarks.ai), CC BY 4.0; fractions converted to percentages; source documents retain their own terms (licence)