Cursor model selection
Use enabled model aliases and a deliberately scoped editor credential.
No current prepared translation is available. The English source is shown without machine translation.
Configure Cursor's OpenAI-compatible custom endpoint with https://api.b2b-llm.com/v1 and a dedicated Model API credential. Select an enabled model or alias that the editor accepts. Do not invent an alias to work around the model picker.
Available aliases
The table is generated from the deployment's enabled model aliases:
| Alias | Model |
|---|---|
ch-46 | claude-opus-4-6 |
ch-46s | claude-sonnet-4-6 |
ch-47 | claude-opus-4-7 |
ch-48 | claude-opus-4-8 |
ch-g54 | gpt-5.4 |
Workspace rollout
Test the selected alias through the account Playground, then enter it into the editor's model configuration. Restart the editor if saved provider settings are not reflected in the model picker. Context, tool and vision options must match the selected model's actual capabilities.
An endpoint override changes the chat/agent model route only where the client supports that setting. It does not establish that every editor feature, completion service or hosted index uses your company endpoint. Retain the editor's required authentication and plan controls.
If a request fails, first compare the full destination path and chosen model with a successful minimal API request. A 401 concerns credentials; an unknown-model response concerns catalog/alias configuration. Use local interception only for an authorized fixed-endpoint client when native configuration is unavailable.