4o Image API

Purpose

4o Image API is an access service for GPT-Image-1 (also known as ChatGPT 4o Image) provided by Kie.ai, aimed at developers who need to integrate the OpenAI 4o image generation model into their own applications. It focuses on "affordable, scalable" API access and is suitable for image generation and editing needs in scenarios such as design, marketing, and creative automation.

Main Capabilities

• Text-to-image and image-to-image: Supports generating high-resolution images from text prompts, and also supports refining existing images through intelligent editing and variation features.
• Accurate text rendering within images: Can clearly present text content such as slogans, UI elements, and product labels in generated images, while ensuring correct placement within context.
• Precise instruction following: Can understand complex prompts and subtle instructions, and recognize the relationships between objects, lighting, and composition, making generated results more aligned with user intent.
• World knowledge and context awareness: Based on OpenAI's multimodal foundation, recognizes real-world objects, cultural elements, and scene logic, generating natural and contextually appropriate visual content.
• Diverse artistic styles: Supports a variety of artistic directions from photorealistic rendering to Ghibli-style anime illustration, with adjustable tones, lighting, and visual aesthetics.
• Character and style consistency: Maintains character and style consistency across multiple generations, suitable for brand avatars, series characters, or continuous scene design.

How to Use

1. Get an API Key: Log in to your Kie.ai account and go to the API Key management page to generate a key. All requests must carry this key in the Authorization header as a Bearer token. Be sure to keep it confidential, and reset it immediately if leaked.
2. Send an image generation request: Send a POST request to the 4o Image Generation API endpoint, with parameters including prompt, size (1:1, 3:2, 2:3), and nVariants (1, 2, or 4). Text-to-image and image-to-image use the same endpoint POST /api/v1/jobs/createTask; you can send a prompt, up to 5 reference images, or both at the same time.
3. Check task status or use callbacks: After submission, a task ID is returned. You can check progress through the /record-info endpoint, or receive automatic updates through the callbackUrl parameter. The response includes generation status, progress, and the final image once the task is complete URL.

Reference images should be passed through files_url; the single-file file_url field is deprecated. Reference images must be publicly accessible URLs. Files that are not yet hosted can first be uploaded using the Kie.ai File Upload API (supports Base64, file streams, or remote URLs). Generated images are stored for 14 days and then expire. It is recommended to use callBackUrl to receive results, and to verify each callback according to the HMAC signature in the Kie.ai webhook guide. Authoritative results can be queried through Get Task Details. New users can get 80 free credits for testing.

Applicable Scenarios

• Ghibli-style art: Generate images with the iconic style of Ghibli animation based on descriptive prompts, for concept art and creative projects.
• Product visualization and display: Generate realistic product models and display images based on text descriptions, without the need for physical prototypes.
• Information and infographic design: Use deep contextual understanding to generate infographics or business charts that clearly convey complex data.
• Character and asset design: Maintain character and style consistency for game art or series creation.

Limitations

• Generated images are stored for only 14 days and cannot be accessed after expiration.
• Reference images must be publicly accessible URLs.
• Specific pricing details, the complete parameter list of available sizes, and commercial use terms are not fully provided in the materials and must be checked in the official pricing card and API documentation.