List your company

Medical Reasoning on MedMCQA

86AccuracyReConcile
36.649.42562.2575.075Jul 14, 2025Sep 10, 2025Nov 7, 2025Jan 4, 2026Mar 3, 2026Apr 30, 2026Jun 28, 2026

Evaluation results

MethodLinks
ReConcileBase Model=GPT-4o2025.0886
MedAgentsBase Model=GPT-4o2025.0885.7
TMA-AllComponBase Model=GPT-4o2025.0885.4
DyLANBase Model=GPT-4o2025.0884.6
MDAgentsBase Model=GPT-4o2025.0880.8
GPT-4.1Access=Close-source2025.0780.58
GPT-4oAccess=Close-source2025.0776.17
Claude-3.5-SonnetAccess=Close-source2025.0776
SMCSCategory=Ours2025.0775.75
HuatuoGPT-o1-72BAccess=Open-source, Parameters=72B2025.0775.25
Open-source Upper BoundCategory=Baselines2025.0775.25
GPT-o3-miniAccess=Close-source2025.0774.92
Symbolic-MoE*Category=Other Methods2025.0774.88
Self-MoACategory=Other Methods2025.0774.88
Self Consistency (Best on Validation)Category=Other Methods2025.0774.83
Claude-3.7-SonnetAccess=Close-source2025.0774.75
Majority VotingCategory=Other Methods2025.0773.33
MoACategory=Other Methods2025.0773.08
DeepSeek-R1-Distill-Llama-70BAccess=Open-source, Parameters=70B2025.0772.5
Simple RouterCategory=Other Methods2025.0772.5
Llama-3.3-70B-InstructAccess=Open-source, Parameters=70B2025.0772.25
Qwen3-32BAccess=Open-source, Parameters=32B2025.0772.17
Llama-3.3-Nemotron-Super-49B-v1Access=Open-source, Parameters=49B2025.0770.92
GLM-Z1-32B-0414Access=Open-source, Parameters=32B2025.0770.5
QwQ-32BAccess=Open-source, Parameters=32B2025.0769.83
Qwen-2.5-72B-InstructAccess=Open-source, Parameters=72B2025.0769.08
DeepSeek-R1-Distill-Qwen-32BAccess=Open-source, Parameters=32B2025.0767.25
Gemma-3-27b-itAccess=Open-source, Parameters=27B2025.0764.58
TMA-AllComponBase Model=MedGemma-4B2025.0863.6
Qwen2.5-32b-InstructAccess=Open-source, Parameters=32B2025.0762.92
Relational AggregationCategory=Ours2026.0562.83
Anchor (Refined Belief)Category=Ours2026.0561.12
GoAMaxCategory=Multi-Agent2026.0560.04
Anchor (Initial Belief)Category=Ours2026.0559.23
EXAONE-Deep-32BAccess=Open-source, Parameters=32B2025.0759.17
Majority (Refined Belief)Category=Ours2026.0558.87
Middle PerplexityModel=Qwen2.5-7B-Instruct, Training=Finetuning, Sample size=1000 examples2026.0658.5
OursModel=Qwen2.5-7B-Instruct, Training=Finetuning, Sample size=1000 examples2026.0658.1
S2LModel=Qwen2.5-7B-Instruct, Training=Finetuning, Sample size=1000 examples2026.0658
GoAMeanCategory=Multi-Agent2026.0557.92
RandomModel=Qwen2.5-7B-Instruct, Training=Finetuning, Sample size=1000 examples2026.0657.7
Embedding DiversityModel=Qwen2.5-7B-Instruct, Training=Finetuning, Sample size=1000 examples2026.0657.5
Majority (Initial Belief)Category=Ours2026.0557.47
m1kModel=Qwen2.5-7B-Instruct, Training=Finetuning, Sample size=1000 examples2026.0657.3
Qwen2.5-Coder-32B-InstructAccess=Open-source, Parameters=32B2025.0757.25
LearnabilityModel=Qwen2.5-7B-Instruct, Training=Finetuning, Sample size=1000 examples2026.0657.1
TeleChat2-35B-32KAccess=Open-source, Parameters=35B2025.0757.08
HIPPOTraining Paradigm=Ours2026.0657
SP3FTraining Paradigm=Reward-Shaping2026.0656
SCCategory=Multi-Agent2026.0555.7
Self-MoACategory=Multi-Agent2026.0555.56
GeneralCategory=Single-Agent2026.0555.22
Qwen2.5_7BTraining Paradigm=Standard Training, Configuration=+ RL2026.0655.1
RefineCategory=Multi-Agent2026.0554.94
MoACategory=Multi-Agent2026.0554.94
ReConcileCategory=Multi-Agent2026.0554.6
Qwen2.5_7BTraining Paradigm=Standard Training2026.0653.8
TMA-AllComponBase Model=Gemma-3-4B2025.0853.3
DebateCategory=Multi-Agent2026.0553.05
DyLANBase Model=Gemma-3-4B2025.0852.4
ReConcileBase Model=Gemma-3-4B2025.0852
PRM RLTraining Paradigm=Reward-Shaping2026.0652
InternLM2.5-20B-ChatAccess=Open-source, Parameters=20B2025.0751.92
PREF-GRPOTraining Paradigm=Reward-Shaping2026.0651.8
MedAgentsBase Model=Gemma-3-4B2025.0850.8
Qwen2.5_7BTraining Paradigm=Standard Training, Configuration=+ SFT2026.0648.4
BiomedicalCategory=Single-Agent2026.0547
CodeCategory=Single-Agent2026.0545.57
MathCategory=Single-Agent2026.0545.25
FinanceCategory=Single-Agent2026.0542.08
LegalCategory=Single-Agent2026.0541.5
MDAgentsBase Model=Gemma-3-4B2025.0838.5