GPT Image 2
GPT Image 2 is OpenAI's earlier image-generation and editing API model. It remains a documented request ID and dated snapshot, but OpenAI introduced GPT-Image-2.5 Flare and Sunburst as the newer API choices in September 2026.
Existing applications may still pin GPT Image 2, while new work needs a deliberate migration decision between Flare and Sunburst. Keeping the predecessor separate prevents the newer ChatGPT product name, model IDs, prices, and provider quality claims from being transferred to an older API contract.
OpenAI Developers still lists `gpt-image-2` and snapshot `gpt-image-2-2026-04-21` for text-and-image input with image output. OpenAI's September 8 launch names GPT Image 2 as the comparison baseline for the new Flare and Sunburst models. Current OpenAI surfaces conflict on relative token pricing, so this page does not claim an exact percentage price difference.
- Generate high-quality images from text prompts.
- Edit or transform images in API-driven creative workflows.
- Maintain an existing integration pinned to `gpt-image-2` or its dated snapshot.
- Run a matched migration evaluation against Flare and Sunburst.
- Compare image model options by quality, latency, token use, cost, and output control.
OpenAI introduced GPT-Image-2.5 Flare as the default for most new applications and Sunburst for higher-precision generation and editing. GPT Image 2 remains a distinct API model rather than becoming an alias for either 2.5 model.
- Undated request model ID: `gpt-image-2`.
- Pinned snapshot: `gpt-image-2-2026-04-21`.
- Text is input only; images are accepted as input and produced as output.
- Generation and edit endpoints are supported; streaming, function calling, structured output, and fine-tuning are not.
Keep GPT Image 2 when an existing application depends on its established behavior and a migration has not passed acceptance checks. Evaluate Flare and Sunburst with the same prompts, reference images, sizes, quality settings, edit sequence, latency target, token use, and human-review criteria before switching.
The new 2.5 model pages say their token rates match GPT Image 2, while OpenAI's general pricing table separately shows lower GPT Image 2 rows than the explicit 2.5 rows. That first-party conflict makes an exact relative-price claim unsafe. Use the active pricing table and billing console for the model and request shape you will actually run.
GPT Image 2 FAQ
Common questions about GPT Image 2.
Is GPT Image 2 an API model or only a ChatGPT feature?+
GPT Image 2 is an API model with request ID `gpt-image-2`. ChatGPT Images 2.5 is the newer product release, and its API choices use the separate Flare and Sunburst model IDs.
Should I migrate from GPT Image 2 to Flare or Sunburst?+
Test Flare first when speed and everyday generation matter, and test Sunburst when edit precision matters most. Keep GPT Image 2 until the replacement passes the same prompts, source images, edit sequence, latency, cost, and review criteria.
Is GPT Image 2 cheaper than the 2.5 models?+
OpenAI's current first-party pages conflict: the 2.5 model pages say token rates match GPT Image 2, while the general pricing table shows different rows. Verify the live pricing and billing console instead of relying on an inferred percentage difference.