Artificial intelligence is reshaping healthcare at an unprecedented pace. Now, OpenAI claims its latest ChatGPT model has surpassed physician-written responses on several clinical tasks, marking a major milestone in healthcare AI.
The announcement accompanies the launch of ChatGPT for Clinicians, a specialized AI assistant designed to help healthcare professionals with consultations, documentation, and medical research. According to OpenAI, the new model achieved higher scores than physician-written answers in the company’s HealthBench Professional benchmark. However, experts continue to stress that AI should support doctors rather than replace them.
OpenAI Unveils a New Healthcare Milestone
OpenAI recently introduced ChatGPT for Clinicians, a free AI tool created specifically for healthcare professionals. The company says the platform was developed with input from hundreds of physicians and focuses on real-world clinical workflows.
Alongside the launch, OpenAI released HealthBench Professional, an open benchmark that evaluates AI systems across three major clinical areas:
- Clinical consultations
- Writing and documentation
- Medical research
The benchmark uses physician-authored conversations and multi-stage reviews by medical experts. Importantly, about one-third of the test cases were created through deliberate “red teaming,” where doctors actively tried to expose weaknesses in the models.
How HealthBench Professional Measures Performance
GPT-5.4 Outperforms Physician Responses
According to OpenAI, GPT-5.4 running inside ChatGPT for Clinicians achieved an overall HealthBench Professional score of 59.0.
In comparison:
- Physician-written responses scored 43.7
- Base GPT-5.4 scored 48.1
- Anthropic Claude Opus 4.7 scored 47.0
- Google Gemini 3.1 Pro scored 43.8
- xAI Grok 4.2 scored 36.1
These results suggest that OpenAI’s healthcare-focused model performs better across several clinical tasks, particularly in writing and documentation.
Benchmark Design Raises Important Questions
Despite the impressive numbers, experts point out that OpenAI designed the benchmark and tested its own models on it. Therefore, independent evaluations will remain essential to validate the results.
Nevertheless, HealthBench Professional is publicly available and aims to become an industry standard for measuring AI performance in healthcare.
Why Clinicians Are Embracing AI
Healthcare professionals are adopting AI faster than many expected.
OpenAI says millions of clinicians already use ChatGPT in clinical practice. In fact, usage among healthcare providers has more than doubled over the past year.
Doctors use AI for many everyday tasks, including:
- Drafting referral letters
- Preparing prior authorizations
- Summarizing patient records
- Reviewing medical literature
- Supporting clinical decision-making
Moreover, several major healthcare organizations have begun integrating OpenAI’s healthcare tools into their workflows. The company believes AI can reduce administrative burdens and allow physicians to spend more time with patients.
ChatGPT Shows Growing Strength in Health Advice
OpenAI’s healthcare ambitions extend beyond clinicians.
The company reports that more than 230 million people use ChatGPT for health and wellness advice every week. Researchers say GPT-5 is the first OpenAI model family developed with healthcare performance as a core objective throughout training.
Earlier studies also found promising results. A study published in 2023 showed that healthcare professionals preferred ChatGPT’s answers to physician responses in 79% of patient questions. Additionally, ChatGPT’s responses were rated as more empathetic and more acceptable overall.
Furthermore, ChatGPT has demonstrated strong performance on several medical examinations and clinical reasoning tests, highlighting its growing capabilities in healthcare settings.
Concerns and Limitations Remain
Despite rapid progress, experts caution against relying entirely on AI for medical advice.
Recent analyses reveal that ChatGPT still struggles with certain situations. For example, it may fail to ask enough follow-up questions, misjudge urgency, or overlook emotional distress in patients. In some cases, researchers found that the model produced concerning recommendations or failed to recognize serious health risks.
Similarly, earlier research found that ChatGPT correctly diagnosed medical cases only about half the time when tested on complex case studies. These findings reinforce the importance of physician oversight.
As a result, most healthcare experts agree that AI should serve as a supportive tool rather than an independent medical authority.
The Future of AI in Healthcare
The healthcare industry is entering a new era where AI and clinicians work side by side.
OpenAI continues to invest heavily in healthcare innovation. The company is developing systems that can understand patient context, analyze health records, and provide personalized assistance while maintaining safety and privacy standards.
Meanwhile, healthcare organizations are exploring how AI can improve efficiency, reduce burnout, and enhance patient care.
Although challenges remain, the latest benchmark results indicate that AI is becoming an increasingly valuable partner in medicine. The focus now shifts toward ensuring these tools remain safe, transparent, and clinically reliable.
Conclusion
OpenAI’s latest ChatGPT model represents a significant advancement in healthcare AI. By outperforming physician-written responses on the HealthBench Professional benchmark, the company has demonstrated the growing potential of AI in clinical settings.
However, AI is not a replacement for medical professionals. Instead, it is emerging as a powerful assistant that can help clinicians work more efficiently and make healthcare more accessible. As technology evolves, the partnership between physicians and AI could redefine the future of medicine.
