I’m seeing what appears to be a GPT-5.3-Codex-Spark entitlement/UI mismatch.
My Usage page shows a dedicated GPT-5.3-Codex-Spark allowance with:
- 100% remaining on the 5-hour limit
- 86% remaining on the weekly limit
However, when attempting to use Spark, Codex returns:
The ‘gpt-5.3-codex-spark’ model is not supported when using Codex with a ChatGPT account.
So the account still displays Spark capacity remaining while Codex explicitly rejects Spark because the session is authenticated through a ChatGPT account.
Expected: Provisioned Spark quota corresponds to usable Spark access, or the Usage page clearly explains that the quota is unavailable.
Actual: Quota is visible with capacity remaining, but the model cannot be selected or used.
Has GPT-5.3-Codex-Spark been intentionally removed from ChatGPT-account access, or is this an entitlement/UI synchronization issue?
I have also reported this through the OpenAI Support chat.
