NativeMind

NativeMind is a local ChatGPT alternative for people who want AI help with reading and writing inside their browser. It processes prompts and page content on your device using local models.

• Chat across tabs: Add open webpages to a conversation to ask questions, summarize content or compare information across sites.
• File analysis: Ask questions about local PDFs, uploaded images and screenshots without sending them to a cloud model.
• Agent mode: Delegate multi-step tasks that combine file analysis and browser-based web search, which doesn't require an external AI API but does need internet access to retrieve web results.
• Writing and email: Rewrite text, proofread, adjust tone and brainstorm, with a Gmail assistant for summarizing messages and drafting replies.
• Page translation: Translate whole pages while preserving their layout, or translate selected text with a side-by-side view.

The Chrome extension is free and open source under AGPL-3.0, and it doesn't require an account. Its Ollama backend supports models including gpt-oss, DeepSeek, Qwen, Llama, Gemma and Mistral. Local model processing works offline once you've downloaded a model. An optional WebLLM demo runs Qwen3-0.6B directly in the browser through WebAssembly.

For developers, an in-browser API lets web apps send prompts to a local LLM and receive its responses.