GPT Image 1

GPT Image 1 is the model that carried OpenAI's image generation out of ChatGPT and into the API on 23 April 2025. The generator had launched inside ChatGPT a month earlier and drew more than 700 million images in its first week; gpt-image-1 made the same natively multimodal system available to developers, and for a year it was the default way to put OpenAI image generation into a product. Technically it is a language model that happens to output pictures: it accepts text and images and returns an image, which is what gave it its two selling points over diffusion-only generators — it follows detailed written instructions, and it renders legible text inside the picture. It offered three fixed shapes (1024×1024, 1024×1536 and 1536×1024), three quality levels and, unlike its successor, transparent backgrounds. Billing was per picture rather than per token: from USD 0.011 for a low-quality square to USD 0.25 for a high-quality landscape or portrait image, with text input at USD 5 and output at USD 40 per million tokens. OpenAI deprecated the model on 22 April 2026, a day after releasing GPT Image 2, and set its shutdown for 23 October 2026; the recommended replacement is GPT Image 2. Its smaller relatives — gpt-image-1-mini and gpt-image-1.5 — were deprecated separately on 2 June 2026 with a shutdown date of 1 December 2026.

▸Videos

No videos yet.

▸Reviews