FLUX Request Moderated vs Content Moderated, and Sume's reason

On BFL's API, Request Moderated means your prompt or image was flagged before processing, and Content Moderated means the generated output was flagged afterward. Both now return a Moderation Reasons array in details. Sume documents no two-stage split and no reasons array: policy refusals can surface under the content_policy_rejected reason, and failed generations are not billed.

BFL facts are from its errors page and release notes; Sume facts are from Image generation and the OpenAPI document, read 2026-10-01.

What are the two BFL statuses?

In both cases result is null, and details names categories such as Violence, Sexual Content or Self Harm.

What does a Sume job report?

Sume's OpenAPI describes public_reason as a machine-friendly string that SDKs and agents can group on, with examples generation_rejected, image_content_rejected and content_policy_rejected. The code that sets it emits content_policy_rejected for policy refusals. This post does not claim Sume returns a category array like BFL's Moderation Reasons.

Am I billed for a moderated request?

The image docs say failed or cancelled generations are not billed, and failed requests return 502 Bad Gateway. Completed generations are billed at endpoint pricing. See the billing section of Image generation for the full list.

How should my client handle it?

Branch on public_reason. Treat content_policy_rejected as a prompt or input problem and change it before resubmitting. For a comparable case on another model, read Seedream 5.0 content policy API errors.