Why is calling the vllm inference model API not recorded in usage?

[Huyueeer]

My proxy contains the Qwen model of Azure OpenAI and VLLM inference. I found that the usage of Azure OpenAI will be recorded in the usage, but the usage of VLLM's Qwen will not be recorded in the usage.

[ishaan-jaff]

Have you set custom pricing @Huyueeer for vLLM ?

https://docs.litellm.ai/docs/proxy/custom_pricing#usage-with-openai-proxy-server-1

[Huyueeer]

Thanks, I've seen the usage.