Hello Aidan !
Thank you for posting on Microsoft Learn.
I think you are experiencing a delay or lack of token-level metric emission for certain configurations or request patterns in Azure OpenAI.
Keep in mind that not all requests emit Total Tokens as only requests that involve completions (chat/completions or completions endpoints) emit token data.
Embedding or moderation calls may show request counts but not token data.
If you're using tools like the playground or embedding APIs, token usage may not be tracked yet in the same way.
Don't forget that Azure may delay token metric publication by up to 15-30 minutes so if you're checking immediately after sending requests, try selecting a wider historical time range.