DALL-E Image Generation Failed

Search your code or automation settings for dall-e-2 and dall-e-3. Replace the value with gpt-image-1-mini (budget), gpt-image-1 or gpt-image-2. Review size and quality values, since GPT Image models let you set quality, size, output format and compression. Check how you read the response: if your code expected a hosted image URL, confirm the format the new model returns and update the code that saves or displays the image. Run a test with a simple prompt such as "a red apple on a white table, studio lighting" before switching production traffic. If a third-party tool has not been updated since May 2026, check for an update or a model setting; otherwise it will keep failing.

Do not retry moderation_blocked automatically without changing the prompt or input images, because the same request is blocked again. Check moderation_stage: input means rewrite the prompt or replace the image; output means aim the prompt at a different result. Remove real people's names and describe a fictional person instead. Replace trademarked characters or logos with original descriptions. Tone down violent, medical or adult detail. Split a complex prompt into a simpler first image, then edit it in a follow-up. The moderation parameter adjusts strictness within policy; it does not allow content the policy forbids.

If you hit an image limit, wait for the reset time shown in the message. If the prompt was filtered, rephrase it. Check status.openai.com for degraded image generation. Refresh the page, start a new chat, disable extensions that alter pages, or try the mobile app. Shorten very long prompts to subject, setting, style and lighting, then add detail in a follow-up.

Check usage and billing in the OpenAI developer dashboard and add credits or raise your spend limit if needed. Add exponential backoff with jitter and retry only on 429 and 5xx responses, never on moderation_blocked. Lower concurrency by queueing image jobs. Cache results so you do not regenerate the same image.

Open the organization settings in the OpenAI developer console and complete verification. It can take some time to take effect, so wait before retrying. Apps that serve individual end users should also send a stable, privacy-preserving safety_identifier with each image request.