List your company
ResearchFormerly WizWand

Multiple-choice Question Answering on MedMCQA

88.9AccuracyGPT-5
-3.50847220.48218944.4728568.463511Feb 23, 2026Mar 7, 2026Mar 20, 2026Apr 2, 2026Apr 15, 2026Apr 28, 2026May 11, 2026
Updated 5mo ago

Evaluation results

MethodLinks
GPT-5Family=Frontier API2026.0588.9
GPT-4.1Family=Frontier API2026.0583.9
GPT-4.1-miniFamily=Frontier API2026.0578.4
S-DPOModel=Llama32026.0572.19
MASS-DPOModel=Llama32026.0571.29
DPO-kModel=Llama32026.0571.04
CLR-voyance-8BFamily=8B family, Merge Strategy=DELLA-Linear2026.0567
Qwen3-8B baseFamily=8B family2026.0566
HuatuoGPT-o1Family=Medical / open-source2026.0564.5
MG-BFamily=4B family, Merge Strategy=Breadcrumbs2026.0561
CLR-voyance-4BFamily=4B family, Merge Strategy=DELLA-Linear2026.0558.5
MG-Δ+AFamily=4B family, Merge Strategy=DELLA-Linear+AIM2026.0558.5
MG-B+AFamily=4B family, Merge Strategy=Breadcrumbs+AIM2026.0558.5
MASS-DPOModel=Qwen32026.0556.66
DPO-kModel=Qwen32026.0555.56
MG-4B baseFamily=4B family2026.0554.5
S-DPOModel=Qwen32026.0552.56
DPOModel=Llama32026.0552.25
S-DPOModel=SmolLM32026.0544.99
Gemma-3-4BFamily=Medical / open-source2026.0544.5
MASS-DPOModel=SmolLM32026.0544.19
DPO-kModel=SmolLM32026.0544.09
DPOModel=Qwen32026.0543.49
Nemotron-CC_ASI+Pre-training Curation Strategy=AI Discovered, Model Parameters=3B, Training Token Budget=500B tokens2026.0340.97
DS-R1-0528Family=Medical / open-source2026.0536
DS-R1-DistillFamily=Medical / open-source2026.0533.5
DPOModel=SmolLM32026.0533.27
MedGemma-27BFamily=Medical / open-source2026.0533
Nemotron-CC_ASIPre-training Curation Strategy=AI Discovered, Model Parameters=3B, Training Token Budget=500B tokens2026.0330.28
DMPOModel=Qwen32026.0528.91
Nemotron-CCPre-training Curation Strategy=AI Discovered, Model Parameters=3B, Training Token Budget=500B tokens2026.0328.86
DCLMPre-training Curation Strategy=Human Discovered, Model Parameters=3B, Training Token Budget=500B tokens2026.0328.15
Fineweb-EduPre-training Curation Strategy=Human Discovered, Model Parameters=3B, Training Token Budget=500B tokens2026.0325.8
DMPOModel=Llama32026.0525.7
DMPOModel=SmolLM32026.0525.5
Ultra-FinewebPre-training Curation Strategy=Human Discovered, Model Parameters=3B, Training Token Budget=500B tokens2026.0324.92
GPT-4o2026.020.5785
Mistral Large2026.020.4344
Llama3-70B-InstructSize=70B, Type=Instruct2026.020.4167
Claude 3 Haiku2026.020.4057
Qwen1.5 (1.8B)Size=1.8B2026.020.047
Gemma (2B)Size=2B2026.020.0457