import type React from 'react'; import { Info, Cpu, Code, BookOpen, HardDrive, Share2 } from 'lucide-react'; export const AboutPage: React.FC = () => { return (

About Antigravity AI Coder

Learn about the architecture and technologies running under the hood.

Hybrid Local/Cloud Architecture

Antigravity Coder is built on a cascading inference pipeline. It is configured to run the lightweight SmolLM2-360M-Instruct model locally in GGUF format using llama-cpp-python.

If the application is deployed on cloud hosts with restricted memory boundaries (e.g. Render Free, Railway Starter), the backend automatically and transparently switches to the Hugging Face Serverless Inference API, guaranteeing high-performance execution without needing a local GPU or extensive memory budgets.

SmolLM2 Engine

Leverages the SmolLM2 360M Instruct model, optimized for fast inference on CPU with excellent instruction-following and code assistance capabilities.

GGUF Quantization

Quantized in Q4_K_M (4-bit quantization). Reduces the weights size to under 400MB and fits comfortably within 1.5GB system RAM, providing blazing-fast local processing speed.

Monaco Code Editor

Integrates the VS Code core editor (Monaco), delivering built-in code formatting, syntax highlighting, search/replace, and advanced editing features directly inside the Playground.

Cloud Container Ready

Dockerized with independent backend and frontend packages, fully prepared for automated pipeline deployment on Railway, Render, and Fly.io.

Antigravity AI Coding Assistant © 2026. Made with Google DeepMind Advanced Agentic Coding.
); };