What you need to know about ChatGPT-5: the main features and feedback from the first users
GPT-5 has become significantly smarter in all areas, as confirmed by its results in academic tests and evaluations involving experts, especially in mathematics, programming, visual perception, and medicine. The model has set a new quality level in mathematics (94.6% on AIME 2025 without the use of tools), applied programming (74.9% on SWE-bench Verified, 88% on Aider Polyglot), multimodal understanding (84.2% on MMMU), and medicine (46.2% on HealthBench Hard) — and these improvements are noticeable in everyday tasks. Thanks to the enhanced reasoning mode in GPT-5 Pro, the model has also set a new record on the GPQA test, scoring 88.4% without the use of tools.
