> ## Documentation Index
> Fetch the complete documentation index at: https://docs.openference.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Chinese LLM Models

> Use GLM, DeepSeek, Kimi, and Qwen through Openference with any supported CLI.

# Chinese LLM Models

Openference routes Chinese models through the same proxy as Western models. Most clients work out of the box once you use the **exact model ID** from `GET /v1/models`.

## Model IDs

Copy IDs exactly — some names include spaces or version suffixes:

| Example ID          | Notes                         |
| ------------------- | ----------------------------- |
| `GLM-5.2`           | GLM                           |
| `DeepSeek-V4-Pro`   | DeepSeek — hyphens, no spaces |
| `DeepSeek-V4-Flash` | Fast, lightweight DeepSeek    |
| `Kimi K2.6`         | Space in the name             |
| `Qwen3.7 Plus`      | Qwen                          |
| `MiniMax-M2.5`      | MiniMax                       |

The catalog changes over time — always confirm against the [Models page](https://openference.com/models) or `GET /v1/models`.

List your available models:

```bash theme={null}
curl https://api.openference.com/v1/models \
  -H "Authorization: Bearer $OPENFERENCE_API_KEY"
```

## Which endpoint each client uses

| Client                              | Endpoint                    | Chinese models               |
| ----------------------------------- | --------------------------- | ---------------------------- |
| **Cursor**, **Cline**, **OpenCode** | `POST /v1/chat/completions` | Supported                    |
| **Claude Code**                     | `POST /v1/messages`         | Supported (Anthropic format) |
| **Codex CLI**                       | `POST /v1/responses`        | Supported                    |

Codex sends Responses API requests. Openference translates `/v1/responses` requests automatically for models whose native format is chat completions, so Codex works with Chinese models out of the box.

## Quick tests

**Chat (Cursor / Cline / OpenCode):**

```bash theme={null}
curl https://api.openference.com/v1/chat/completions \
  -H "Authorization: Bearer $OPENFERENCE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"GLM-5.2","messages":[{"role":"user","content":"hi"}],"max_tokens":5}'
```

**Messages (Claude Code):**

```bash theme={null}
curl https://api.openference.com/v1/messages \
  -H "Authorization: Bearer $OPENFERENCE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"GLM-5.2","max_tokens":10,"messages":[{"role":"user","content":"hi"}]}'
```

**Responses (Codex):**

```bash theme={null}
curl https://api.openference.com/v1/responses \
  -H "Authorization: Bearer $OPENFERENCE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"GLM-5.2","input":"hi","max_output_tokens":5}'
```

## Troubleshooting

* **404 model not configured** — Check spelling and spaces against `/v1/models`.
* **403 model not allowed** — Your API key may have model restrictions.
* **529 Overloaded** — Temporary capacity issue; does not consume quota. Retry or switch model.

See also:

* [Codex CLI](/integrations/codex)
* [Cursor](/integrations/cursor)
* [Claude Code](/integrations/claude-code)
* [Compatibility Matrix](/integrations/compatibility-matrix)