### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

I wanted to integrate Claude CLI using the LiteLLM proxy, so that requests in the Anthropic format could be forwarded to the Zhipu large model for processing. However, this integration wasn’t possible. Upon further investigation, it turned out that LiteLLM tried to access OpenAI’s /responses endpoint. Unfortunately, Zhipu’s OpenAI-compatible interface didn’t support this endpoint, resulting in a 404 error. Even enabling the “Compatible with Anthropic” settings didn’t resolve the issue.

### Steps to Reproduce

1. Build the Litellm UI panel and create a glm-5.2 model compatible with OpenAI. Configure the corresponding API_baseurl and API key. 2. Tests in the UI panel’s Playground, as well as curl simulations for /v1/message and /v4/reponses, all report the same error. 3. The curl request is as follows: curl -X POST http://0.0.0.0:4000/v1/messages -H "Authorization: Bearer $LITELLM_MASTER_KEY" \ -H "Content-Type: application/json" -d ‘{ "model": "GLM-5.2" "max_tokens": 1000, "messages": [{"role": "user", "content": "Testing response: OK"}] }'

##…