Skip to content

Benchmark

Vals MedCode

Scores are percentages. Higher is better.

Models measured
99
Scores recorded
197
Sources
2
IQ weight
1.68

Score distribution

42%
Mean
42.4%
Median
9%
Std dev
31.3%
10th percentile
53.2%
90th percentile
19.7%
Min
19.7%63.6%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.68 for chat models, measured over 99 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
101GLM 5.1clinical-result-2003· Clinical Benchmarks data docs41.6%
102DeepSeek V4 Flash 0731· Vals AI41.4%
103DeepSeek V4 Flash 0731clinical-result-2004· Clinical Benchmarks data docs41.4%
104Claude Opus 4.1clinical-result-2005· Clinical Benchmarks data docs41.4%
105GPT 5.4· Vals AI41.3%
106GPT 5.4clinical-result-2006· Clinical Benchmarks data docs41.3%
107Inkling FP4· Vals AI41.2%
108Inkling FP4clinical-result-2007· Clinical Benchmarks data docs41.2%
109DeepSeek V4.1 Flash· Vals AI41.2%
110DeepSeek V4.1 Flashclinical-result-2008· Clinical Benchmarks data docs41.2%
111MiMo V2.6 Flashclinical-result-2009· Clinical Benchmarks data docs41.1%
112MiMo V2.6 Flash· Vals AI41.1%
113GPT 5.4 Nanoclinical-result-2010· Clinical Benchmarks data docs41%
114GPT 5.4 Nano· Vals AI41%
115GLM 5.2· Vals AI40.8%
116GLM 5.2clinical-result-2011· Clinical Benchmarks data docs40.8%
117Mistral Large 4· Vals AI40.7%
118Qwen3.8 Maxclinical-result-2012· Clinical Benchmarks data docs40.7%
119Qwen3.8 Max· Vals AI40.7%
120Claude Sonnet 4.5clinical-result-2013· Clinical Benchmarks data docs40.6%
121gemini-2.5-flash-preview-09-2025· Vals AI40.5%
122DeepSeek V4 Pro· Vals AI40.5%
123DeepSeek V4 Proclinical-result-2015· Clinical Benchmarks data docs40.5%
124Gemini 2.5 Flash· Vals AI40.4%
125gemini-2.5-flash-preview-09-2025thinking· Vals AI40.3%
126Kimi K2.6· Vals AI40.1%
127Kimi K2.6clinical-result-2018· Clinical Benchmarks data docs40.1%
128Kimi K2.5clinical-result-2019· Clinical Benchmarks data docs39.3%
129Kimi K2.5thinking· Vals AI39.3%
130Qwen3.7 Max· Vals AI38.8%
131Qwen3.7 Maxclinical-result-2020· Clinical Benchmarks data docs38.8%
132Nemotron 3 Ultra· Vals AI38.6%
133Nemotron 3 Ultraclinical-result-2021· Clinical Benchmarks data docs38.6%
134Grok 4clinical-result-2023· Clinical Benchmarks data docs38.1%
135Grok 4· Vals AI38.1%
136Grok 4.3clinical-result-2024· Clinical Benchmarks data docs38.1%
137Grok 4.3· Vals AI38.1%
138Inkling Small· Vals AI37.9%
139Inkling Smallclinical-result-2025· Clinical Benchmarks data docs37.9%
140Grok 4 Fast· Vals AI37.4%
141Grok 4 Fastclinical-result-2026, reasoning· Clinical Benchmarks data docs37.4%
142Qwen3.6 Plus· Vals AI36.9%
143Qwen3.6 Plusclinical-result-2027· Clinical Benchmarks data docs36.9%
144Llama 4 Maverick· Vals AI36.5%
145Llama 4 Maverickclinical-result-2028· Clinical Benchmarks data docs36.5%
146Claude Sonnet 4clinical-result-2029· Clinical Benchmarks data docs35%
147Claude Sonnet 4thinking· Vals AI35%
148MiniMax M2.7· Vals AI34.4%
149MiniMax M2.7clinical-result-2030· Clinical Benchmarks data docs34.4%
150Gemini 2.5 Flash Lite Preview 09-2025thinking· Vals AI34.2%

Showing 101–150 of 197 results.

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