### What happened?

From time to time, when making requests with a virtual API key under certain circumstances, liteLLM enters an endless loop of trying that request, failing, then trying again, despite the client only having made that request once.

This can happen, at least, when one of these conditions are met, although I haven't been able to reliably reproduce the issue on these conditions alone:

* The virtual API key is over budget. * The virtual API key is over the TPM limit * The model name is incorrect (specifically a leading capital letter "Model-abcd" vs the existing "model-abcd")

In all of these cases, the logs are a constant stream of the following exception (Copy pasted an exception for the third case mentioned above, changing the actual model for 'Model1'.

``` 14:53:33 - LiteLLM Proxy:ERROR: common_request_processing.py:721 - litellm.proxy.proxy_server._handle_llm_api_exception(): Exception occured - 400: {'error': '/chat/completions: Invalid model name passed in model=Model1. Call `/v1/models` to view available models for your key.'} Traceback (most recent call last): File "/usr/lib/python3.13/site-packages/litellm/proxy/proxy_server.py", line 4803, in chat_c…