veta.Docs Projects
DocsArtificial intelligence

Local models

Prepare the server

  1. Download and load a model in LM Studio or Ollama.
  2. Start its OpenAI-compatible server.
  3. Allow connections from Veta's origin in the server's CORS settings.

See the current options in LM Studio or Ollama.

Connect from Veta

  1. In Providers, choose the LM Studio or Ollama preset.
  2. Enter the server address. localhost means the device running your browser.
  3. Save the connection. In Models → Add model, choose from the list or enter its identifier.

Compatible provider configuration

If it does not connect

  • Another device: use its network address, not localhost.
  • Local network permission: grant it when your browser requests it for this connection.
  • Mixed content: an HTTPS page may block an HTTP API. Use HTTPS with a valid certificate.
  • CORS: the server must allow Veta's exact origin.

Requests go directly from the browser. Veta does not install a proxy or change your network. Connection tests.