Premium requests consumed very fast if using litellm

[chenyanming]

Very fast, it is like single prompt consume 30. However, copilot-api consume 1 to 2. May I know why?

But copilot-api got me error sometimes:
⎿  API Error: 466 {"error":{"message":"bad request: the specified API version is no longer supported. You may need to update your client to a newer version.\n","type":"error"}}

[feiskyer]

this is caused by a known bug in litellm, you can follow BerriAI/litellm#18155