localAI

Use local AI within the browser

What is LocalAI?

LocalAI is designed to bridge the gap where traditional tools fall short—offering access to web content, local files, and on-the-go resource fetching.

Why use it?

If you value privacy but still want the power of LLMs, LocalAI is for you. There’s no need to install applications, deal with configurations, or handle version compatibility issues. It works seamlessly with Chrome, Firefox, and their derivatives.

Key Benefits:

• Local Control: Your data stays on your device—no sharing or storing of your actions. You control what to keep or delete.
• Task Assistance: Need summaries or quick insights? LocalAI helps with tasks using the information in your active tab.
• Web Search: Ask the AI to search the web and it will use your configured search engine in a background tab — no external APIs, no credential sharing.
• Read Links (opt-in): The AI can open a link in a separate background tab and read it, leaving your current tab untouched. It asks first in a Local AI pop-up window (Allow always, Allow this time, Cancel) unless you turn on Allow navigation in the options; the tab is closed when the answer is finished. See Allow navigation.
• Copy as Markdown: Copy any AI reply as its original Markdown, including replies restored from history.
• Custom Prompts: Create, store, and execute your own prompts locally for a personalised experience.
• Backup and Transfer: Export your settings, user prompts and functions to one file with Export All in the options and restore them with Import All. See Export All / Import All.
• External Hooks: Enhance functionality by integrating your own solutions and resources.

What do you need to use it?

You’ll need enough hardware power, preferably with a GPU.

A local AI tool—these are easy to set up, usually just download and run:

• Ollama — the recommended runtime, used as the API endpoint.

Ollama is local-first: everything runs on your machine, and the same endpoint can serve cloud models when you need more, without reconfiguring the extension or storing API keys in it.

If you use Ollama Cloud, the preferred setup is still the local Ollama endpoint. Sign in with the Ollama app or CLI first, then keep the extension pointed at your local /api/* endpoint. Cloud-tagged models exposed by the signed-in local daemon can then be used without storing API keys in the extension.

The model picker loads two catalogues:

• Local models are fetched from your configured Ollama endpoint /api/tags.
• Cloud models are fetched from https://ollama.com/api/tags.

If Ollama Cloud access is a concern in your environment, note that refreshing the model catalogue will make that external request for the cloud tab.

The same local daemon can back other AI tools you already use — Claude Code (ollama launch claude) and Codex CLI (ollama launch codex, or codex --oss) — so one local setup serves all of them.

Note

Other local runtimes

Ollama is the default choice. LM Studio and Mozilla LlamaFile are the alternative local runtimes this project has worked with.

And the browser extension.

It’s easy—give it a try and let us know your thoughts!

Where to get it?

	Browser	Extension
	Chrome	link
	Firefox	link

---

Important

If you are a Firefox user, please read this.

---

Here is the documentation where you can get a deeper understanding.