> ## Documentation Index
> Fetch the complete documentation index at: https://docs.4096bytes.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Cursor

> Configure a 4096bytes custom model in Cursor

Cursor can connect to 4096bytes through OpenAI compatible settings. After configuration, you can select a 4096bytes model in Cursor Chat or Agent.

<Warning>
  Cursor custom API Keys usually affect only the selected Chat or Agent model. Cursor Tab, completion, indexing, and background tasks may still use Cursor services. The exact scope depends on your Cursor version.
</Warning>

## Prerequisites

* You have completed [Create an account](/en/quickstart/create-account).
* You have completed [Get an API Key](/en/quickstart/get-api-key).
* Cursor is installed and **Settings** opens normally.
* You have the 4096bytes API Key, Base URL, and model name.

Create a dedicated API Key for Cursor, such as `cursor-desktop` or `cursor-work`.

## Install

If Cursor is not installed, open the [Cursor download page](https://cursor.com/downloads) and choose the package for macOS, Windows, or Linux.

<Steps>
  <Step title="Download the installer">
    Choose the correct package for your machine. On macOS, distinguish Apple Silicon from Intel. On Windows, distinguish x64 from ARM64.
  </Step>

  <Step title="Launch Cursor">
    Open Cursor once after installation and confirm the editor starts normally.
  </Step>

  <Step title="Open settings">
    Open **Settings** and confirm you can see model, AI, or provider-related settings before continuing.
  </Step>
</Steps>

## Open model settings

<Steps>
  <Step title="Open Cursor settings">
    Open **Settings** in Cursor. Common entry points include the lower-left gear icon, **Cursor Settings** in the command palette, or **Cursor > Settings** in the menu.
  </Step>

  <Step title="Open the model page">
    Open **Models**, **AI**, or a provider-related page. Find OpenAI API Key, custom API, or Override Base URL settings.
  </Step>

  <Step title="Enable OpenAI compatible settings">
    If there is a switch, enable OpenAI API Key or custom OpenAI settings. Fill 4096bytes as an OpenAI compatible provider.
  </Step>
</Steps>

## Fill connection fields

| Cursor field                           | Value                                               |
| -------------------------------------- | --------------------------------------------------- |
| `OpenAI API Key`, `API Key`            | The API Key you created in the 4096bytes dashboard. |
| `Override OpenAI Base URL`, `Base URL` | `https://api.4096bytes.com`.                        |
| `Model name`, `Custom model`           | The model name shown in the dashboard.              |

Use the root Base URL only. Do not append `/v1/chat/completions`, `/v1/responses`, or another full endpoint path.

<Warning>
  The API Key is saved in local Cursor settings. Do not configure a personal key on shared computers, and do not share screenshots that expose the key.
</Warning>

## Add the model name

<Steps>
  <Step title="Add a custom model">
    Click **Add model**, **Add custom model**, or a similar action in the Cursor model list.
  </Step>

  <Step title="Enter the model name">
    Enter the model name shown in the dashboard. It must match exactly, including capitalization, hyphens, and version numbers.
  </Step>

  <Step title="Save settings">
    Save the API Key, Base URL, and model name. If Cursor offers **Verify** or a test button, run it.
  </Step>
</Steps>

## Select and test the model

After saving, open Cursor Chat or Agent and choose the 4096bytes model from the model selector.

Use a low-risk test prompt:

```text theme={null}
Describe the purpose of the current project in one sentence. Do not modify any files.
```

If the model responds, Cursor can send requests through 4096bytes.

## Tips

* Use a dedicated API Key for Cursor.
* Start with a short prompt to confirm the model, key, and Base URL.
* If multiple providers are configured, check the selected model before sending a message.
* For teams, use separate keys per person or device to diagnose usage.

## Troubleshooting

<AccordionGroup>
  <Accordion title="The new model does not appear">
    Confirm the model was added to the model list and restart Cursor. Some versions refresh the selector only after reopening the Chat panel.
  </Accordion>

  <Accordion title="model not found">
    Copy the model name again from the dashboard and check for extra spaces, capitalization errors, or missing version suffixes.
  </Accordion>

  <Accordion title="Authentication failed">
    Check that the API Key is complete. Make sure you did not paste an old, deleted, or newline-containing key.
  </Accordion>

  <Accordion title="Wrong request address">
    Use `https://api.4096bytes.com` as the Base URL. Do not fill a full endpoint path or copy another client's address format.
  </Accordion>

  <Accordion title="Cursor still uses official models">
    Manually choose the 4096bytes model in Chat or Agent. Custom API Keys may not affect every built-in Cursor feature.
  </Accordion>

  <Accordion title="Requests fail even though settings look correct">
    Check account balance, key quota limits, model availability in the current group, and network access to `https://api.4096bytes.com`.
  </Accordion>
</AccordionGroup>