OpenAI gpt-image-1

The premier natively multimodal model from OpenAI: it generates, edits, and modifies high-fidelity images from both text and image inputs.

This is OpenAI's most sophisticated image generator, a major upgrade from the DALL-E series. Codenamed 'gpt-image-1,' it operates on an autoregressive architecture—a key difference from typical diffusion models—and consistently tops the Artificial Analysis Image Arena leaderboard. It excels at complex tasks like precise image editing and consistent text rendering within the visual output. Following its deployment in ChatGPT, the model saw explosive adoption, with 130 million users creating over 700 million images in the first week. Developers access the model via the Images API, enabling image generation up to 1536x1024 resolution with quality tiers (Low, Medium, High) for cost control.

https://platform.openai.com/docs/guides/images