Claude Desktop can now be configured to use Ollama as a third-party gateway provider, letting developers run open models locally or via Ollama's cloud while still keeping the ability to switch back to Anthropic models. Setup is a toggle within Ollama; telemetry is disabled by default and Ollama enforces a zero data retention policy. Users can swap between cloud models for larger tasks and local models configured through Ollama's Settings page, switching back to their previous Anthropic setup at any time.
Questions this post answers
How do I use open models with Claude Desktop instead of Anthropic's models?
Claude Desktop can be configured to use Ollama as a third-party gateway provider. Open Ollama, select Claude, and turn it on; Ollama automatically configures the gateway so you can then choose any local model or a model from Ollama's cloud. Turning the Claude toggle off in Ollama restores the previous Anthropic setup. Anyone weighing local versus cloud models for Claude workflows can follow tool integration updates like this on daily.dev.
Will my prompts be sent to Anthropic or Ollama when using Ollama models inside Claude Desktop?
No, telemetry is disabled by default when connecting Claude Desktop to Ollama. Ollama also applies a strict Zero Data Retention policy across all of its models and services, whether accessed through its cloud or run locally, so prompts and data are not retained or shared with Anthropic. Developers evaluating data privacy tradeoffs across AI tool integrations can track these details via daily.dev.
Can I switch back to Anthropic models after connecting Claude Desktop to Ollama?
Yes, switching between Anthropic models and Ollama is done by toggling the feature inside Ollama. Turning the integration off restores the previous Claude setup, while turning it on lets you use any Ollama model, local or cloud, in place of Anthropic's own models. Those comparing model providers for their Claude setup can follow changes like this on daily.dev.