SynapseModel

This project is a decentralized AI model hub built on Web3 that enables anyone to upload, discover, and interact with AI models in a secure and transparent environment. Users connect their crypto wallets to access the platform, where we have integrated Polygon’s x402 protocol for smart payments. From there, they can explore a wide range of AI models and run them directly within an interactive live playground powered by Fluence.

All model files are stored on Walrus decentralized storage, ensuring permanence and accessibility. The Fluence runtime drives model execution, delivering streaming outputs and configurable parameters. On-chain smart payments via Polygon x402 allow seamless, usage-based transactions with flexible pricing models.

The frontend is built with React + TypeScript, styled with Tailwind CSS, and enhanced by Framer Motion for smooth interactions and a modern user experience. Users can browse models, search by type or cost, view detailed information, and securely transact before launching the playground. An upload portal enables developers to contribute their models, set usage options, and publish them directly to the hub.

Our mission is to establish a trustless, user-centric AI ecosystem that unites the principles of decentralization with the capabilities of modern AI, delivering a next-generation experience for both creators and users.

Our platform is built on a modern, component-based architecture using React with TypeScript, providing robust type safety and an enhanced developer experience. The user interface leverages Tailwind CSS for consistent design patterns and Framer Motion for fluid animations and micro-interactions that enhance user engagement throughout the platform. The frontend follows a modular component architecture with reusable elements including ModelCard, WalletModal, and ChatUI components. This approach ensures code maintainability and consistency across the platform while enabling rapid feature development and maintaining a cohesive user experience. Web3 functionality is implemented through a comprehensive integration stack utilizing wagmi and ethers.js for blockchain interactions and smart contract communication, web3modal for seamless wallet connectivity across multiple providers, and Polygon network as the default chain for cost-effective transactions. The platform integrates the x402 protocol to enable programmable micropayments, allowing creators to monetize model usage while providing users with granular payment control. Model storage and retrieval is powered by Walrus, a decentralized storage network. We developed a lightweight client wrapper that provides progress tracking for upload operations, manifest generation and retrieval, and accessible link management for stored models. Model execution runs on Fluence, a decentralized compute platform, with our custom-built client library handling session initialization and management, real-time output streaming, runtime parameter configuration, and interactive playground functionality for model testing. The codebase maintains high reliability through comprehensive testing using Jest and React Testing Library, focusing on critical user flows and component behavior validation. During the rapid development phase, we implemented a strategic prototyping approach by creating mock implementations for Walrus and Fluence clients to enable parallel development, stubbing payment flows using ethers.js for UI/UX validation, and maintaining production-ready architecture patterns to facilitate seamless integration of live services. This approach allowed for accelerated development while ensuring the underlying architecture could support full production deployment without significant refactoring.