OpenAI has officially launched GPT-5, featuring a 500,000-token context window and significantly improved multimodal reasoning capabilities. The new model can process and analyze combinations of text, images, audio, and video in a single interaction with what the company describes as near-human-level understanding.
Benchmark results show GPT-5 outperforming its predecessor by 34% on complex reasoning tasks and achieving expert-level performance on professional licensing exams across medicine, law, and engineering. The model also demonstrates substantially improved factual accuracy and reduced hallucination rates.
