Fix wrong claims in async queue reference

[Eric-Laurence]

Four things the async queue page claimed that the API does not do.

Cancelling a generation does fire media_generation_completed, carrying status: "cancelled" and only the identifying fields, so a client that reads url or error off every event will raise. Audio models queue like any other and come back as media_type: "audio". response_format is accepted and then discarded by the worker, so asking for b64_json quietly gets you a URL. Running out of credit answers 402 with the shortfall and a count of what you already have queued, which is what anyone batching will hit first.

Each of these was verified against prod rather than read off the code. The error table also stops quoting the num_generations message word for word so it cannot drift from what the API says.

[mintlify]

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated
oxenai	🟢 Ready	View Preview	Sep 18, 2026, 1:59 AM

💡 Tip: Enable Automations to automatically generate PRs for you.