Vercel Fluid: Revolutionizing Serverless with a New Compute Model for APIs, Streaming, and AI

Vercel has launched Vercel Fluid, a novel compute model that allows a single worker to manage multiple requests while preserving the benefits of serverless architecture, thereby enhancing compute efficiency and addressing cold start issues. By prioritizing existing resources over creating new instances, Fluid optimizes resource use for long-running tasks and AI inference and enables a many-to-one architecture capable of managing tens of thousands of concurrent requests on a single function. It offers advantages such as efficient auto-scaling, horizontal and vertical concurrency, and optimized I/O efficiency under a pay-as-you-go pricing structure. Fluid utilizes microVMs more effectively than traditional serverless architectures by allowing a VM to handle multiple concurrent requests, thereby improving efficiency. The Vercel Functions router orchestrates function execution to reduce cold starts and boost concurrency, using persistent TCP tunnels for secure communication and distributing workloads. Fluid compute is designed for modern workloads, dynamically adjusting to traffic while reducing compute costs by up to 85% through prioritized resource use and eliminating idle time. Additionally, Vercel Fluid integrates seamlessly with SSOJet's API-first platform, offering secure Single Sign-On (SSO), Multi-Factor Authentication (MFA), and Passkey authentication for enterprises, enhancing security and optimizing user management processes.