Atlas

Benchmarks

← All benchmarks

MedMCQA

Health & Medicine · 2022-03-27

MedMCQA is a four-option multiple-choice medical question-answering dataset built from Indian postgraduate medical entrance examinations (AIIMS and NEET PG) across 21 subjects. Results are reported on the 4,183-question development split, because test labels are withheld, as answer accuracy; the Open Medical-LLM leaderboard's harness-specific results are kept separately. Higher is better.

Top models (higher is better)

ModelScore
o383.3
Gemini 2.5 Pro81.1
Gemini 2.5 Flash79.7
GPT-4o (source unspecified)76.1
MedGemma 27B Text IT74.2
MedGemma 27B IT70.2
Gemma 3 27B IT62.6
MedGemma 1.5 4B IT59.8
MedGemma 4B IT55.7
Gemma 3 4B IT45.4
Loading Atlas data…