Benchmark Registry

MedXpertQA

MedXpertQA results reported for 41 models across 2 versions. Updated 2026-10-07.

Report an issue

Versions

2 versions, plus 1 variant or configuration

VersionReleasedMetricModelsResults
MedXpertQA MMLatest
1 variant of MM
VariantConfigurationReleasedMetricModelsResults
MedXpertQA MM — Mercor single-shotMercor single-shot agentJanuary 30, 2025Accuracy3335
January 30, 2025Accuracy99
MedXpertQA TextJanuary 30, 2025Text multiple-choice accuracy11

Recently reported results: MedXpertQA MM

Latest version: MedXpertQA MM

ModelScoreSource
Qwen3.8-Max80.4%qwen.aiAccuracyReported Launch postQwen TeamSelf-reportedRecord
Gemma 4 12B Instruct (thinking mode)48.7%huggingface.coAccuracyReported Model cardGoogleSelf-reportedGoogle DeepMindRecord
Gemma 4 E4B Instruct (thinking mode)28.7%huggingface.coAccuracyReported Model cardGoogleSelf-reportedGoogle DeepMindRecord
Gemma 4 E2B Instruct (thinking mode)23.5%huggingface.coAccuracyReported Model cardGoogleSelf-reportedGoogle DeepMindRecord
Gemma 4 26B A4B Instruct (thinking mode)58.1%huggingface.coAccuracyReported Model cardGoogleSelf-reportedGoogle DeepMindRecord