Skip to content
inferamdocs
Website ↗

Editors

Cursor

Use Inferam in Cursor's chat and run the Claude Code extension on Inferam.

  • A paid Cursor plan (Pro or higher) for Cursor’s chat. Cursor only lets paid plans use your own API key: on the free plan, chat stops with “Cursor Subscription Required” while the key is on. This is Cursor’s rule, not Inferam’s.
  • For Claude Code inside Cursor, the Claude Code extension. It doesn’t use Cursor’s models, so it works on any Cursor plan, free included.
  • Cursor installed. You don’t need to have opened it: setup starts it once if it has never run.
  • The Inferam CLI. Setup signs you in if this computer isn’t signed in yet.
Terminal window
inferam setup cursor

Open Cursor’s chat: DeepSeek-V4.1-Flash-1017 is already selected. If you use the Claude Code extension, its next session runs on Inferam too.

Cursor keeps its model settings in a database that it rewrites while running, so if Cursor is open, setup quits it, saves the settings and opens it again. Your windows and unsaved edits come back as they were. If you’ve never opened Cursor, setup starts it once so Cursor creates its settings first.

Setup writes these into Cursor’s settings database (User/globalStorage/state.vscdb in Cursor’s folder: ~/Library/Application Support/Cursor on macOS, ~/.config/Cursor on Linux, %APPDATA%\Cursor on Windows). You’ll see them in Cursor Settings → Models:

Setting Value
OpenAI API Key On, with Cursor’s own key from inferam.ai/keys
Override OpenAI Base URL On, https://api.inferam.ai/v1
Models DeepSeek-V4.1-Flash-1017 added, switched on and selected for chat and the other features

The settings these replace are saved in the same database, and --undo puts them back.

In Cursor’s user folder (User/settings.json in the folder above):

{
"claudeCode.claudeProcessWrapper": "/Users/you/.local/bin/inferam",
"claudeCode.disableLoginPrompt": true
}

The extension starts Claude through inferam, exactly as inferam claude does in a terminal, so it runs on Inferam with this computer’s sign-in.

Your Cursor plan is free, and Cursor only lets paid plans use your own API key in chat. Upgrade to Cursor Pro, or run inferam setup cursor --undo to turn the key off and get Cursor’s own models back. --undo also takes the Claude Code extension off Inferam.

Setup waits up to 30 seconds for Cursor to quit. A dialog asking about unsaved work can hold it open; setup then stops without changing anything. Answer the dialog and run inferam setup cursor again.

The model isn’t selected, or the settings look unchanged

Section titled “The model isn’t selected, or the settings look unchanged”

Cursor was probably still running and wrote its own copy back. Run inferam setup cursor again; it quits Cursor before writing.

Run inferam setup cursor again. It replaces Cursor’s key with a new one, which also fixes a key revoked on inferam.ai/keys.

  • Requests go through Cursor’s servers. Cursor builds the final prompt on its servers and sends it to Inferam from there, so Cursor’s privacy policy also applies to what you send. Its zero-data-retention mode doesn’t cover your own keys.
  • Cursor’s other OpenAI models go to Inferam too while the base URL override is on, where they don’t exist. inferam setup cursor --undo turns it off.
  • Agent mode varies by Cursor version. Chat works; some Cursor versions limit your own models in Agent mode.
  • Tab completion always uses Cursor’s own model.
  • On Cursor Teams and Enterprise plans, Cursor charges its own per-token rate on requests that use your own key. Inferam itself is free during the beta.
Terminal window
inferam setup cursor --undo

Puts back the model settings you had before, removes the two Claude Code settings, and revokes Cursor’s key. Like setup, it restarts Cursor if it’s open.