Local-first · v0.12.3
Chat with local LLMs, right in your browser.
Ollama Client is an open-source browser extension for Ollama, LM Studio, llama.cpp, user-added OpenAI-compatible servers, and Anthropic. Chat with local AI models inside Chrome and Firefox with RAG, file upload, web search, tool calling, and page context — all without a mandatory cloud API.
Features
What you can do without leaving your browser.
Multi-provider local models
Verified Ollama, LM Studio, and llama.cpp profiles, plus user-added OpenAI-compatible endpoints and Anthropic.
Multi-chat sessions
Multiple concurrent chats with branch navigation and SQLite-backed local persistence.
Enhanced content extraction
Defuddle integration with smart scroll strategies plus GitHub and YouTube extraction helpers.
Optional web search
Tool-capable models can search through SearXNG, Brave Search, or Tavily while keeping backend choice in settings.
Text-to-speech
Searchable voices, pitch and rate controls, per-message playback. Cross-browser voice loading.
Prompt templates
Pre-built prompts for summarization, translation, and code help. Custom templates supported.
File upload & RAG
PDF, TXT, MD, DOCX support with background embedding progress. Configurable max file size.
Backup & restore
Full data snapshots via ZIP backups with versioned manifests. Restore reports partial failures.
Visible privacy controls
Review local data, backup or wipe it, tag chats, and fork message branches without leaving the extension.
Advanced configuration
Provider base URL control, selected model mapping, capability checks, debug logging.
Privacy first
No mandatory cloud APIs. Local data storage. Open source. You control the endpoint.
Get started
Three steps from install to first chat.
- Step 1
Start a local provider
Run Ollama, LM Studio, or llama.cpp. Note the endpoint URL.
- Step 2
Install the extension
Add it from the Chrome Web Store and pin it for easy access.
- Step 3
Pick a model and chat
Open the side panel, pick a model, chat on any website. CORS issue? The setup guide covers it.
Stack
Built withReact 19 · WXT · TypeScript · shadcn/ui · sql.js · Defuddle.
Endpoint privacy is yours to configure — see theprivacy policyfor what stays local and what leaves.