Research
/
The Consensus Mechanism
Second opinion matters: towards adaptive clinical AI via the consensus of expert model ensemble. How Sully built a multi-expert system that outperforms single frontier models at clinical reasoning.
61.2%
MedXpertQA accuracy
Research
/
Model strengths vary by medical domains
A cross-specialty benchmark analysis showing how leading large language models differ in accuracy, reasoning, and specialization across medical tasks and body systems.
13 of 14
models under 30% on basic science
Research
/
Large-scale analysis of LLMs in healthcare
A continuously evolving benchmark of medical AI performance across clinical reasoning, documentation, and coding tasks.
99.1%
ICD-10 coding accuracy with Sully retrieval
Research
/
Scalable Architecture for Multi-Modal Healthcare AI
Modular, purpose-built agents that share one communication layer and reach clinicians and patients through every channel they already use.
5
channels on one agent layer
Research
Towards Open-Source Foundational Medical Agents
Why Sully runs its medical agents on open-source foundation models, and how ensembling and consensus close the gap with proprietary systems.
98%
adjusted coding accuracy, gpt-oss-120b
Research
Sully.ai’s Doctor-Language Model vs SOTA LLMs
Building safe and ethical AI for healthcare: how Sully.ai’s Doctor-Language Model compares with GPT-4 and GPT-3.5 at turning visit transcripts into SOAP notes.
