Ollama local vs cloud models: what actually leaves your machine?

Updated November 3, 2026.

Ollama can answer a prompt in two places. A local model runs on your own computer. A cloud model runs on Ollama's servers, so the prompt and the response travel to Ollama. Ollama's FAQ puts the difference this way: Ollama runs locally, and the company does not see your prompts or data when you run locally, but when you use cloud-hosted models it processes your prompts and responses to provide the service.

How to tell which one you are using

Ollama's quickstart says to sign in to use cloud models or choose a local model, and that you can use cloud models with an API key or run models locally without one. The cloud page adds a naming detail. In the Ollama app or command line, a cloud model is named with the cloud suffix, for example gemma4:cloud, and cloud models do not need to be downloaded. For API requests sent directly to ollama.com you use the name returned by the tags list instead, such as gemma4:31b. The quickstart's local example is different: ollama run gemma4:e2b downloads the model and starts a chat on your computer.

The API introduction shows the other tell. Direct cloud access uses the base URL https://ollama.com/api and needs an API key, while the local server uses http://localhost:11434/api and local requests do not need a key. It also says that to use cloud models through your local server, you sign in to Ollama. So a machine that is signed in can reach cloud models through the same local address, which is why the model name and the sign-in state matter before you send anything sensitive.

What the documentation says about cloud data

The cloud page states that Ollama processes cloud prompts and responses to answer your requests and does not use them to train models. The FAQ adds that for cloud-hosted models it does not store or log that content, that it collects basic account information and limited usage metadata that does not include prompt or response content, and that it does not sell your data. Those are statements from Ollama's documentation, so read the privacy policy it links to if the exact terms matter to your work.

Keeping a machine local only

Ollama's FAQ describes a local only mode. Set disable_ollama_cloud to true in ~/.ollama/server.json, or set the environment variable OLLAMA_NO_CLOUD=1, then restart Ollama. Once disabled, the logs show Ollama cloud disabled: true. The FAQ warns that turning off the cloud features means you lose the ability to use Ollama's cloud models and web search. The cloud page repeats the instruction to disable cloud features if you want to use only local models.

The same FAQ says the server binds to 127.0.0.1 port 11434 by default and that the bind address changes with the OLLAMA_HOST variable, so check that setting if other devices on your network can reach the server. Our security checklist for a home AI server covers that side. You can see where a loaded model is running with ollama ps; its Processor column shows whether a model sits in CPU memory, on the GPU, or split between them.

When local is the better fit

Local models keep the prompt on hardware you control, but they must fit that hardware. The quickstart's local example, Gemma 4 E2B, has a download of about 7.2 GB, and Ollama recommends 8 GB of available VRAM or unified memory on a Mac, adding that with less VRAM Ollama can use system RAM but responses may be slower. How much memory a given model needs is on our GPU memory page; treat any figure for your own machine as an estimate until you test it.

Frequently asked questions

Does a local Ollama model send my prompts anywhere?

Ollama's FAQ says it does not see your prompts or data when you run locally. Cloud-hosted models are different: Ollama processes those prompts and responses to provide the service.

How do I know a model is a cloud model?

In the Ollama app or command line, cloud models use the cloud suffix, such as gemma4:cloud, and require signing in. Sending API requests directly to ollama.com uses names from the tags list, such as gemma4:31b, plus an API key.

Can I turn the cloud features off?

Yes. Set disable_ollama_cloud to true in ~/.ollama/server.json or set OLLAMA_NO_CLOUD=1, then restart Ollama. You lose cloud models and web search while it is off.

Do local requests need an API key?

No. The API introduction says cloud requests need an API key and local requests do not.

Sources

Sources checked November 3, 2026. Figures that depend on your exact model, runtime or machine are estimates; check the source or test on your own hardware.

Adults only (18+). Private uncensored AI · Terms · Privacy · All guides