Any model
Claude, GPT, Gemini or a model on your own Mac
OpenDot is not tied to one AI company. Use a cloud model, run one locally with Ollama or LM Studio, or point it at any compatible URL, and switch per Dot.
Bring the model you already use
OpenDot supports cloud models from Anthropic (Claude), OpenAI (GPT), Google (Gemini), xAI (Grok), Mistral and DeepSeek, plus aggregators such as OpenRouter and Groq. You paste your own API key, it is stored in the macOS Keychain, and the provider bills you directly. OpenDot itself is free.
Run a model on your Mac with Ollama or LM Studio
For a fully local assistant, run Ollama, LM Studio, llama.cpp or vLLM. OpenDot detects them automatically and lists their models under On this Mac. A Dot running on a local model does its thinking entirely on your machine, with no API key and no per-token cost.
If you just want to try it, start Ollama with a model, for example ollama pull llama3.1, and open OpenDot. Our step-by-step guide to running a local AI assistant on your Mac with Ollama walks through the whole setup.
Any compatible URL works too
Using a company gateway, a self-hosted server or another provider that speaks the OpenAI or Anthropic API format? Add it as a custom provider with its URL. The Test button checks that it responds, and Discover models lists what it offers.
A different model for every Dot
You set a default model in Settings, Models, and any Dot can use a different one. A Dot that triages email can run on a small, fast local model, while your Research Dot uses a larger cloud model. Switching is a setting, not a migration.
Privacy depends on where the model runs
With a cloud model, only what the model needs to answer is sent, and personal details such as emails, phone numbers and card numbers are masked first. With a local model, nothing about the Dot's thinking leaves your Mac. Read more on how OpenDot protects your data.
Choosing between cloud and local
A cloud model is usually the most capable and needs nothing but a key. A local model costs nothing to run and keeps everything on your Mac, at the price of using your Mac's own horsepower. You do not have to choose once for the whole app: put the Dots that handle sensitive mail on a local model and the Dots that do heavy research on a cloud model.
Because the model is a setting on each Dot, you can also change your mind later without recreating anything. Switch a Dot from one provider to another and its name, personality, memory and permissions stay exactly as they were.
Frequently asked questions
Does OpenDot work with Ollama?
Yes. Start Ollama with a model, for example ollama pull llama3.1, and OpenDot finds it automatically. LM Studio, llama.cpp and vLLM are detected the same way.
Do I need an API key?
Only for cloud models. Paste a key from Anthropic, OpenAI, Google, xAI, OpenRouter and others, and that provider bills you directly. Local models need no key and cost nothing.
Can different Dots use different models?
Yes. Set a default in Settings, Models, then pick a different model for any individual Dot.
Can I use my own server or gateway?
Yes. Any OpenAI-compatible or Anthropic-compatible URL can be added as a custom provider.
