Providers in Cue

Choose an installed agent or connect a hosted or local model. Configure it in Settings → Providers, then select it from a chat’s agent menu.

Choose a provider

Cue groups providers by how they run. Choose the agent or model connection that matches your setup.

Installed agents

These agents run their own CLI processes and handle their own sign-in.

  • Claude Code. Runs the installed claude command and uses its sign-in.
  • Codex. Runs the installed codex command and uses its sign-in.
  • Antigravity. Runs the installed agy command. Cue offers Full and Edit Files access modes.

Models through Cue

Cue runs the agent loop with the selected model.

  • OpenRouter. Connect an account or API key to browse hosted models.
  • Local models. Download or import a GGUF model, or use an existing llama.cpp server.
  • Apple Intelligence. Uses Apple’s on-device model when it is available on this Mac.
Cue for Mac Providers settings with Claude Code, Codex, Antigravity, OpenRouter, Local models, and Apple Intelligence in two groups.
Installed agents
Models through Cue
The two groups in Cue’s Providers settings.

Set up a provider

Open Settings → Providers and select the provider’s row for its connection and model controls.

Claude Code, Codex, or Antigravity

Install and sign in through the agent’s own CLI. Select it from the agent menu in Cue. Open its Providers row to adjust any model, thinking, or access options Cue offers for that agent.

OpenRouter

  1. Open Settings → Providers → OpenRouter. Select Connect OpenRouter for the browser connection, or use API key options.
  2. In Model, select a model or choose All Models… to browse. Paid requests use credits on the connected OpenRouter account.
Cue’s OpenRouter settings with account connection and model browsing controls.
Connect account
Browse models
OpenRouter connection and model controls. The screenshot shows an example selection; available models can change.

Local models

  1. Open Settings → Providers → Local models.
  2. Choose Cue on this Mac to download or import a GGUF model. Cue starts its bundled llama.cpp server for that model.
  3. Choose Existing llama.cpp server if you already operate one. Enter its connection details and use Save & Check.
Cue’s Local models settings with Cue on this Mac and Existing llama.cpp server.
Cue on this Mac
Your own server
Two local-model paths in Cue.

Apple Intelligence

Enable Apple Intelligence in macOS System Settings on an eligible Mac and wait for its model to be ready. Then select Apple Intelligence in Cue.