The new model gpt-image-1 is now available for developers for image generation
Following the successful launch of image generation in ChatGPT with over 700 million images created within the first week, the innovative model is now also available via the API. Known as gpt-image-1, it allows developers and companies to seamlessly integrate professional image generation into their tools and platforms. The model is characterised by its versatility, enabling it to generate images in a wide range of styles, accurately follow specific instructions, and even precisely render text.
Cross-industry applications
Numerous leading companies and start-ups are already using image generation in their products. Partners include:
- Adobe: Integration into Firefly and Express enables creative experiments with various aesthetic styles.
- Airtable: Facilitates the management of marketing campaigns through the generation and adaptation of media content.
- Figma: Users can generate and edit images directly within the platform.
- Gamma: Supports the creation of presentations and websites with over 5 million AI images daily.
- HeyGen: Offers enhanced avatar creation and editing.
- OpusClip: Generates appealing thumbnails for over 10 million YouTube creators.
- Quora: Utilises the model as a new standard for image quality and accuracy.
- Wix: Allows users to create and customise professional designs with a few clicks.
- Photoroom: Introduces new AI tools for creating high-quality product images.
- Playground: Provides users with advanced editing options for designs.
Further applications
Additionally, OpenAI is working with more partners on new use cases:
- Canva: Plans integration for improved design and editing features.
- GoDaddy: Experiments with logo creation and the creation of social media and marketing content.
- HubSpot: Exploring applications for creating marketing materials.
- Instacart: Testing the creation of recipe and shopping list images.
- invideo: Enhances video creation with precise text generation and style guidance.
Safety measures
The gpt-image-1 model uses the same safety measures as image generation in ChatGPT. This includes measures to prevent the creation of harmful images and the use of C2PA metadata in generated images. Developers can also adjust moderation sensitivity with a special parameter.
Pricing model
The use of gpt-image-1 is billed per token, with different prices for text and image tokens.
The model is globally available via the API, with upcoming support in the Responses API. Some developers may need to verify their organisation to gain access. Further information is provided in the usage guide, while the detailed article on the subject offers additional background.