Cursor

The custom model entry point in Cursor varies across versions. When you can see settings like OpenAI API Key, Override OpenAI Base URL, or Add Custom Model, you can connect to Tokeness using the OpenAI-compatible method.

If your current version does not offer these entry points, or if third-party models are unstable in Agent tool calls, switch to Cline, Claude Code, Codex CLI, or OpenCode.

Connection details

Field	Value
OpenAI API Key	A key created in the Tokeness console
Override OpenAI Base URL	https://n.tokeness.dev/v1
Custom Model Name	Full model name from the Tokeness model marketplace

The field names in Cursor may still say "OpenAI". Once Base URL override is enabled, requests are sent to the address you specify.

1. Prepare Tokeness

Prepare the following first:

1. A Cursor-dedicated API key.
2. A model name suited for coding tasks.
3. Available balance.

You can name the key:

cursor-local

2. Open Cursor model settings

In Cursor, go to:

Settings -> Models

Find the API Keys, OpenAI API Key, or similar area.

3. Fill in the API key and Base URL

1. Enable OpenAI API Key.
2. Paste your Tokeness API key.
3. Enable Override OpenAI Base URL.
4. Set the Base URL to https://n.tokeness.dev/v1.
5. Save or confirm.

If Cursor shows a confirmation dialog, confirm that it will send OpenAI requests to the custom Base URL before proceeding.

4. Add a custom model

On the Models page, find Add Custom Model.

Enter the model name copied from the Tokeness model marketplace. After adding it, disable the default models you do not need, to avoid selecting the wrong one during testing.

For the first test, use a short prompt:

Please reply with a single sentence confirming Cursor is now connected to Tokeness.

Then check the Tokeness usage logs to confirm the request.

5. Scope of use

Cursor's custom Base URL is typically suited for:

Scenario	Notes
Regular chat	Start with short questions
Single-file Q&A	Control the context range
Small-scale code explanation	Observe whether the model can reply stably

For large-scale Agent edits, complex tool calls, and multi-file modifications, compatibility varies across third-party models. Start with small tasks, then expand the scope.

6. Troubleshooting

Symptom	Fix
Cannot find Override OpenAI Base URL	The current Cursor version may not support this entry point
401 Unauthorized	Re-copy the Tokeness API key
404 Not Found	Set Base URL to https://n.tokeness.dev/v1
model not found	Re-copy the custom model name from the model marketplace
No Tokeness logs	The current model is not using the custom OpenAI configuration
Chat works but Agent errors	The model is not fully compatible with Cursor tool calls; switch models or tools
Still using the default model	Disable default models or manually select the custom model in the chat sidebar

External docs

• OpenRouter Cursor Integration
• SambaNova Cursor Integration
• LiteLLM Cursor Integration