xszwow

xszwow / Chrome2api

Public

OpenAI-compatible local API wrapper for Chrome Gemini Nano / ChromeML

13
6
85% credibility
Found May 30, 2026 at 13 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
C++
AI Summary

Chrome2api is an open-source project that creates a local wrapper around Google's on-device Gemini Nano AI model, which is built into the Chrome browser. It allows users to run the AI directly on their Windows computer without needing Chrome open, without an internet connection, and without sending data to the cloud. The project provides both a command-line tool for quick AI queries and a local web server that mimics the popular OpenAI API format, so existing chat tools can use it. Users must provide their own copy of Chrome's AI model files (extracted from their installed Chrome browser), which the project clearly explains are proprietary Google files and are intentionally not included in the repository.

How It Works

1
🔍 You discover a way to run AI locally

You find Chrome2api, a project that lets you run Google's on-device AI model (Gemini Nano) directly on your Windows computer without needing the Chrome browser open.

2
📦 You gather the required pieces

You copy a few large files from your installed Chrome browser into the project folder—the AI model weights and two supporting libraries that Chrome uses to run AI.

3
🔨 You build the runner program

You run a simple script that compiles the source code into a small executable that can talk to Chrome's AI runtime.

4
⚡ You run the AI directly from your command line

You type a simple command with your prompt, and the AI responds instantly—no internet, no browser, just your local machine thinking.

5
You can use it in two different ways
💻
Direct command line

You run a single command with your prompt and get an immediate text response, with optional support for images or audio.

🌐
Local web API server

You start a local server that works like OpenAI's API, so any chat tool that supports OpenAI can talk to your local AI.

6
💬 You chat with your local AI

You send prompts with text, images, or audio, and the AI responds—everything runs privately on your own computer.

🎉 You have a private, offline AI assistant

You now have a fast, local AI that works without internet, keeps your conversations private, and responds like a cloud AI service.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 13 to 13 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is Chrome2api?

Chrome2api wraps Chrome's built-in Gemini Nano model and exposes it as an OpenAI-compatible HTTP API server running on your local machine. The project provides a bare C++ runner that talks directly to Chrome's internal ML runtime DLLs, then layers a REST API on top so you can query the model through standard chat completion endpoints. You get text generation, image input, and audio transcription through a familiar interface without sending data to any cloud service.

Why is it gaining traction?

The hook is clear: run a local AI model with zero setup beyond having Chrome installed. Because the API speaks OpenAI, it slots into existing toolchains immediately--Dify workflows, LangChain chains, and Copilot integrations all work without modification. The model runs on your GPU or CPU using WebGPU, and the project includes PowerShell scripts for launching the server and running the bare runner directly.

Who should use this?

Windows developers curious about Chrome's on-device AI who want to experiment with local inference. Researchers prototyping multimodal pipelines without cloud costs. Anyone building OpenAI-compatible local LLM applications who specifically wants Chrome's model rather than Ollama or llama.cpp. This is not for production deployments or anyone who needs cross-platform support.

Verdict

Chrome2api scores 0.8500000238418579% on credibility--a polite way of saying this is highly experimental. With 13 stars and no pre-built binaries, you must extract proprietary Chrome DLLs, compile the C++ runner with Zig, and manually wire up model weights that exceed GitHub's file limits. The documentation is thorough for what it covers, but the setup friction is substantial. Worth exploring if you want to tinker with Chrome's on-device model outside the browser, but treat it as a proof-of-concept, not a library.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.