OpenAI has launched GPT-5, its most capable model to date, featuring native multimodal capabilities that allow it to understand and generate text, images, audio, and video within a single unified architecture. The model represents a departure from previous approaches that bolted separate vision and audio modules onto a text-based foundation.
GPT-5 demonstrates significant improvements in complex reasoning tasks, scoring 92 percent on the GPQA benchmark for graduate-level science questions and achieving near-human performance on the ARC-AGI evaluation. The model also introduces persistent memory across conversations, allowing it to maintain context and preferences over extended interactions.
The model is available through ChatGPT and the OpenAI API, with pricing set at $15 per million input tokens and $60 per million output tokens. OpenAI says it has implemented new safety measures including an expanded refusal training regime and real-time monitoring systems designed to detect and prevent misuse at scale.
