Skip to main content

OpenAI GPT Image 2 – editing

Prompt-driven editing of existing images, with the same three quality tiers as GPT Image 2 generation and its own pricing.
GPT Image 2 Edit is an image editing API that modifies one or more images you supply according to a text instruction. It is the reference-image counterpart of GPT Image 2: same model, same three quality tiers, same batch limit, but the input is an existing image plus a prompt rather than a prompt alone. Use it for retouching, replacing or removing objects, restyling a scene, and for correcting or replacing text that already appears in the image.

​
Key capabilities

  • Prompt-driven editing: describe the change in natural language instead of masking it by hand
  • Three quality tiers: quality accepts low, medium and high. The tier changes both the render quality and the credit cost per output image — editing is priced differently from generation, see Pricing below
  • Batch variants: request up to 10 edited variants of the same input in one call with num_images. Each output image is billed
  • In-image text edits: legible text rendering carries over to editing, so labels and copy inside the image can be corrected or replaced
  • Async processing: every request returns a task ID immediately. Poll the task endpoint or supply webhook_url to be notified on completion

​
Use cases

  • Product photography retouching: change backgrounds, lighting or props without a reshoot
  • Localisation: replace on-image copy for another market while keeping the layout
  • Packaging revisions: update a label, claim or price on an approved key visual
  • Creative variations: generate up to 10 alternative treatments of one approved image for A/B testing
  • Clean-up: remove unwanted objects, watermarks or distractions from an existing asset

​
Edit images with GPT Image 2

Submit the source image or images plus an instruction to create an editing task. The API responds with a task ID; collect the result by polling or via webhook.

POST /v1/ai/text-to-image/gpt-image-2-edit

Create a new image editing task

GET /v1/ai/text-to-image/gpt-image-2-edit

List all GPT Image 2 Edit tasks with status

GET /v1/ai/text-to-image/gpt-image-2-edit/{task-id}

Get task status and results by ID

​
Parameters

The source images are supplied on the request body alongside the prompt. The endpoint reference pages above are generated from the OpenAPI specification and are the authoritative, complete parameter list, including the exact field and accepted formats for the input images.

​
Pricing

Editing sends your source images to the model, and their input tokens dominate the cost, so it is not priced like a generation of the same size: resolution does not affect the price of an edit — the input images set the cost and the output size barely moves it. Every image in num_images is billed.

​
Frequently Asked Questions

GPT Image 2 Edit is an image editing API powered by OpenAI’s GPT Image 2 model. You submit one or more source images together with a text instruction and receive a task ID immediately. Poll the task endpoint or configure a webhook to receive the edited images when processing completes.
No. The edit is described in the prompt rather than painted as a mask. Be specific about what should change and what must stay the same, since the model decides the region itself.
quality selects one of three tiers — low, medium or high. A higher tier spends more compute per output image, which improves detail and text fidelity, and costs more credits per image: a high edit costs six times a low one, and medium four times.
Up to 10, via num_images. Every output image is billed at the tier you selected, so a request for 10 variants at high costs ten times a single high image.
GPT Image 2 generates a new image from a text prompt alone. GPT Image 2 Edit starts from images you provide and modifies them. Both use the same model, the same three quality tiers and the same batch limit, and both bill per output image.
Rate limits depend on your subscription tier. See Rate Limits for current limits.
Cost per output image depends on the quality tier you request. See the Pricing page for current rates and subscription options.

​
Best practices

  • Name what stays: state explicitly which parts of the image must be preserved, not only what should change
  • One change at a time: a single, clearly scoped edit per request is more reliable than a list of unrelated changes
  • Tier the workflow: explore variants at quality: "low", then re-run the chosen instruction at quality: "high"
  • Batch deliberately: num_images is billed per output image, so batch at low and finalise at high
  • Production integration: use webhook_url instead of polling for scalable applications
  • Error handling: implement retry logic with exponential backoff for 503 errors