### Check for existing issues

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

### What happened?

Using Z.AI Coding Plan through LiteLLM Proxy on the documented coding endpoint, plain `glm-5.2` works but the documented 1M variant `glm-5.2[1m]` fails with `Unknown Model`.

Z.AI's Coding Plan docs say to use `glm-5.2[1m]` for 1M context on `https://api.z.ai/api/coding/paas/v4`.

In practice, LiteLLM Proxy returns:

```json {"error":{"message":"litellm.BadRequestError: ZaiException - Unknown Model, please check the model code.. Received Model Group=zai/glm-5.2[1m]\nAvailable Model Group Fallbacks=None","code":"400"}} ```

I reproduced this both by:

1. sending `model: "zai/glm-5.2[1m]"` directly through the proxy, and 2. exposing a local alias `glm-5.2-1m` mapped to `zai/glm-5.2[1m]` in `config.yaml`.

What I expected to happen:

- LiteLLM Proxy should successfully pass through `zai/glm-5.2[1m]`, or - LiteLLM's Z.AI provider/docs should explicitly document that the `[1m]` suffix is not valid for proxy usage on the OpenAI-compatible Coding Plan endpoint.

Related context:

- LiteLLM issue `#25479` already documents a separate Z.AI Coding Plan m…