OpenAI's GPT-5.5 Instant Surpasses Doctors in Medical Knowledge Evaluations

Key Highlights

Table of Contents

ChatGPT attracts 230 million weekly users seeking health and wellness guidance
Former Google Med-PaLM architect Karan Singhal now spearheads OpenAI’s medical AI initiatives
GPT-5.5 Instant achieved superior scores compared to doctor-authored responses in communication and precision evaluations
Medical response factual errors declined 71% over an eight-week period based on real-world usage data
The ChatGPT Health feature, debuted in January with medical record integration capabilities, maintains a waiting list for access

Weekly healthcare queries on ChatGPT now exceed 230 million, according to OpenAI’s latest disclosure. Users rely on the platform for diverse medical needs, from interpreting laboratory reports to planning physician consultations.

Healthcare has emerged as a strategic priority for the company, with researcher Karan Singhal at the helm of this initiative.

Improving human health will be one of the most personal, tangible impacts of AGI.
As our models continue to improve, our goal is to make ChatGPT more accurate, more useful, and more impactful in those moments — and to keep bringing that progress to more people.…
— OpenAI (@OpenAI) June 18, 2026

Karan Singhal’s Background and Role

Singhal transitioned to OpenAI during the middle of 2024 following his tenure at Google, where he served as a key architect of Med-PaLM, specialized artificial intelligence systems engineered for medical inquiries. Google subsequently reduced its commitment to that initiative, pivoting toward broader AI applications.

Upon joining OpenAI’s ranks, Singhal established a dedicated healthcare research division and collaborated with over 200 practicing physicians. This partnership aimed to infuse genuine medical expertise into ChatGPT’s health-related responses.

His team developed HealthBench, a specialized assessment framework created in collaboration with medical professionals to evaluate and benchmark AI healthcare capabilities.

GPT-5.5 Instant’s Performance Metrics

OpenAI’s newest freely available model, GPT-5.5 Instant, underwent comparative testing against responses written by physicians and earlier AI versions. The model demonstrated superior performance across multiple dimensions: precision, communication quality, thoroughness, and utility for health-related decision-making.

Medical professionals evaluating the system identified fewer critical shortcomings. These deficiencies typically include overlooking urgent warning signs, neglecting to request additional patient context, or failing to recommend professional medical consultation when appropriate.

According to OpenAI’s analysis of billions of weekly health-related conversations, the frequency of flagged factual inaccuracies decreased by 71% during the previous two-month period.

The GPT-5 series represents OpenAI’s first model family intentionally optimized for healthcare guidance throughout every development phase.

Enhanced Patient Context Understanding

AI-powered health tools face a fundamental limitation: minimal patient background information. Traditional physicians access comprehensive medical histories and records, while conversational AI typically operates without such context.

OpenAI addressed this gap in January by introducing a specialized ChatGPT health feature that integrates with wellness applications and accepts uploaded medical documentation. Singhal illustrated this capability by describing how he uploaded sleep tracking information from his Apple Watch, enabling the system to determine that bedroom temperature was compromising his deep sleep quality.

Access to ChatGPT Health remains restricted via a waiting list more than five months following its announcement.

Singhal emphasized the team’s commitment to improving the chatbot’s questioning methodology to mirror physician consultation patterns before providing recommendations. He noted that medical professionals are rapidly adopting AI technologies, with institutional resistance proving less significant than many anticipated.

Ongoing Challenges and Legal Scrutiny

OpenAI continues defending itself against litigation claiming that GPT-4o, a previous model version, provided dangerous medical guidance to users. The company has rejected these allegations. Legal proceedings remain active.

Nevertheless, OpenAI has accelerated healthcare feature expansion rather than retreating from this domain.

GPT-5.5 Instant is now accessible to all free-tier ChatGPT users, democratizing access to enhanced health responses without requiring premium subscriptions.