Last month, OpenAI introduced a major update to ChatGPT's image generation capabilities, powered by a new model. The update went viral, and as a result, over 130 million users created more than 700 million images in just one week.
Now, OpenAI is bringing the updated image generation model to developers via the gpt-image-1 API. With this new API, developers will be able to generate high-quality images within their applications and services. This new gpt-image-1 API supports various visual styles, precise image editing, rich world knowledge, and the ability to render text accurately.
Right now, gpt-image-1 is available via the Images API. OpenAI is working on support for the Responses API, and it will be coming soon.
Developers can control moderation sensitivity with the moderation parameter. By default, standard filtering will be enabled. However, developers can specify low for less restrictive filtering. Developers can also control the image quality, the quantity of images generated, whether the background of images should be transparent or opaque, and the output format, such as JPEG, PNG, and WebP.
The image generation API is priced per token, with different pricing for text and image tokens. For text input, it costs $5 per 1M tokens. For image input, it costs $10 per 1M tokens. For image output, it costs $40 per 1M tokens. In a real-life scenario, developers would pay around $0.02, $0.04, and $0.19 per image for low, medium, and high-quality square sizes, respectively.
Interested developers can now explore this model in the OpenAI Playground. Several companies, including Adobe, Airtable, Figma, Gamma, HeyGen, Wix, Photoroom, Canva, GoDaddy, HubSpot, Instacart, Invideo, and OpusClip, are already using OpenAI’s image generation API in their products and experiences.
0 Comments - Add comment