May 2025 Summaries

The text discusses the limitations of traditional serverless computing in handling Large Language Models (LLMs) and their interactions, which require sustained compute resources and continuous execution patterns. A new compute model called Fluid is introduced, designed to address these challenges by prioritizing existing resources before spawning new ones, scaling inside a function, and dynamically reallocating compute where needed. This approach reduces overhead, enables efficient scaling, and ensures that every function invocation actively contributes to processing. Additionally, Fluid compute provides edge security with Vercel Firewall, secure instance architecture, enhanced reliability and availability, and is optimized for AI workloads requiring efficiency and security.

Vercel has successfully blocked over 148 billion malicious requests from 108 million unique IP addresses since February, cutting response time and expanding control. The company's real-time DDoS filtering, managed Web Application Firewall (WAF), and enhanced visibility ensure consistent, proactive security for its users. Vercel has also introduced new security enhancements such as the Bot Filter, which challenges non-browser bot traffic without disrupting critical automations. Additionally, the company has rolled out faster denial-of-service mitigation with Protectd, a powerful update to its denial-of-service (DoS) mitigation infrastructure. Furthermore, Vercel has released a comprehensive Secure SDLC whitepaper and proactively hardened its platform against high-severity vulnerabilities in open-source frameworks. The company also continues to improve its Firewall usability, including finer control during active threats and deeper visibility into malicious traffic patterns.

Vercel Blob is a durable storage solution integrated with Vercel's application delivery network, providing simple and globally available storage that feels native to your app. It's built for high-volume workloads, leveraging AWS S3 infrastructure for 99.999999999% reliability, and introduces Blob Data Transfer, a new high-volume delivery strategy separate from Fast Data Transfer. Caching is abstracted away by Vercel's caching infrastructure, handling data transfer as part of the standard request lifecycle. The solution supports large uploads up to 5TB with built-in support for multi-part uploads, resumability, retries, and concurrency. Pricing is usage-based and in line with what developers already understand, with regional pricing for Pro and Enterprise customers. Vercel Blob also introduces a new Observability tab for team-level visibility into how Blob stores are used across applications, allowing developers to identify inefficiencies and optimize their workflow.

Vercel recently published a Model Context Protocol (MCP) adapter that makes it easy to spin up an MCP server on most major frameworks, enabling users like Vapi to build real-time voice agents. Vapi's MCP server was rebuilt using the Vercel MCP adapter, allowing them to support both Server-Sent Events and Streamable HTTP transports out of the box, and leveraging Fluid Compute for optimized resource utilization. By deploying their MCP server on Vercel, Vapi can focus on building their product instead of managing infrastructure, with features like observability and OAuth being added as their server grows.

The Vercel AI Gateway is now available for alpha testing, allowing developers to switch between ~100 AI models without managing API keys, rate limits, or provider accounts. Built on the AI SDK 5 alpha, the Gateway handles authentication, usage tracking, and will eventually include billing. The current speed of AI development requires frequent updates to model APIs in application code, which can be frustrating for developers. The Gateway aims to help ship fast and keep up with AI progress without needing multiple API keys and provider accounts. It also addresses production capacity issues by load balancing across providers and providing failover options. The Gateway is built on top of the AI SDK, which is free and open source, and will initially offer usage for free with rate limits based on Vercel plan tier, with plans for pay-as-you-go pricing when it reaches general availability.

The Vercel AI Gateway, now available for alpha testing, aims to streamline the use of AI models by allowing seamless switching between approximately 100 models without the hassle of managing API keys, rate limits, or provider accounts. Built on the AI SDK 5 alpha, the Gateway handles authentication, usage tracking, and plans to integrate billing in the future, offering developers flexibility in navigating the fast-paced AI landscape. It addresses common issues in production AI applications, such as capacity and rate limiting, by enabling load balancing and failover across providers, thus ensuring consistent performance and availability. The AI SDK, which the Gateway builds upon, provides a unified abstraction for AI model APIs across various modalities and is free and open-source. During the alpha phase, usage of the AI Gateway is free with certain rate limits, and future plans include pay-as-you-go pricing and possible support for bring-your-own-key options. Although the Gateway is not yet ready for production use, it is designed to leverage cost savings and performance improvements as model inference costs decrease and providers compete on quality and price.

Fern, a multi-tenant documentation platform, uses Vercel and Next.js to achieve efficient multi-tenancy, faster development cycles, and significantly improved load times. By running on Vercel's infrastructure, Fern benefits from automatic caching, optimized content delivery, and instant scalability, which has driven a 50-80% reduction in page load times. The platform's architecture is built around a single Next.js app running on Vercel, handling multi-tenant routing across various customer domains. With Vercel's custom domain support, Fern supports both subdomains and subpaths, enabling customers to bring their own domains with just a CNAME DNS record. Fern also leverages Vercel Functions for rapid iteration, cutting deploy times down to five minutes and gaining the ability to test changes in preview environments before shipping to production. Additionally, Vercel's monorepo support allows Fern to deploy updates across all customer sites in minutes without managing infrastructure overhead. The platform has achieved enterprise-grade performance with fast-loading documentation, reduced latency through global content delivery, and automatic caching and Incremental Static Regeneration. Fern recently migrated 65% of its platform from Pages Router to Next.js App Router, unlocking major performance gains. With 63% month-over-month growth, Fern is rapidly scaling to serve customers worldwide, delivering a high-performance, multi-tenant documentation experience without the complexity of managing traditional infrastructure.

MetaMask, a widely used self-custodial wallet, was migrated to a new architecture by Consensys, leveraging Next.js and Vercel to improve operational reliability, scalability, and iteration. The migration aimed to address limitations of the previous Static Site Generation approach, enabling faster workflows, dynamic content, and fewer bottlenecks. With the new setup, the team achieved significant performance improvements, including improved uptime, accelerated deployments, and enhanced SEO outcomes. The architecture now supports features like Incremental Static Regeneration and smart caching, allowing for richer dynamic content and real-time collaboration across teams. Consensys is confident in this setup to scale with their vision, enabling teams to focus on innovation rather than infrastructure.

The Vercel team is updating their `v0` service to use a metered pricing model based on input and output tokens, which convert to credits. This change aims to provide more predictable pricing as users grow and increases the amount of usage available on their free tier. Existing `v0` users will transition to the new pricing at the start of their next billing period, while new users will start using the improved pricing today. The updated model includes included monthly credits for each plan type, with varying amounts based on plan level. Users can track their usage and remaining credits in their settings page, and team owners can monitor usage across their entire team. The change also introduces a more dynamic pricing structure, where different actions cost varying amounts based on input and output tokens.

The Vercel Open Source Program is a developer initiative that provides resources, credits, and support to open source project maintainers. The program aims to help communities thrive by giving them the tools they need to ship faster and scale confidently. This spring's cohort of projects includes Vecto3d, Script Kit, Lenis, Fumadocs, Shadcn-table, Morphic, Zola, Valibot, shadcn-form, jsPDF, LobeChat, Milkdown, KokonutUI, nuqs, Nemo, RetroUI, Tweakcn, CourseLit, Comp AI, EsJS, Gridiron Survivor, and Frontpage. These projects are gaining popularity in their respective communities, with some having over 40K site visits or 1 million weekly npm downloads. The program welcomes a new cohort of projects four times a year, providing them with $3,600 in Vercel credits, perks from partners, and a dedicated Slack space to learn from one another.

The latest updates from Vercel aim to deepen the integration of feature flags into developers' workflows, making it easier to experiment and iterate with confidence. The Flags Explorer allows users to manage feature flags directly from the Vercel Toolbar, simplifying testing and experimentation. New integrations with flag providers on the Vercel Marketplace enable seamless deployment of experimentation tools without requiring separate accounts. The Flags SDK provides a flexible way to control features at runtime, allowing developers to import and evaluate flags using the library. By integrating these updates, users can accelerate iteration velocity, reduce risk, and deliver personalized user experiences with confidence.

The text discusses updates to Vercel's feature flag management and experimentation tools, making it easier for developers to integrate these features into their workflow. The Flags Explorer in the Vercel Toolbar allows users to manage feature flags directly from the toolbar, simplifying testing and experimentation. New integrations with flag providers on the Vercel Marketplace enable quick setup and managed configuration of experimentation platforms like Statsig. The Flags SDK provides an open-source library for integrating feature flags into Next.js and SvelteKit applications, allowing developers to control features at runtime without redeploying. These updates aim to accelerate iteration velocity, reduce risk, and deliver personalized user experiences, empowering teams to build more quickly and confidently with Vercel's DX platform.

Vercel has introduced several updates to enhance feature flag integration within workflows, including the general availability of the Flags Explorer, which allows users to manage feature flags directly from the Vercel Toolbar. This feature enables users to test and override feature flags without affecting others, facilitating smoother experimentation and feedback collection. Additionally, Vercel's Marketplace now includes a new Experimentation category, featuring Statsig, which allows for quick setup and consistent deployments without needing separate accounts. The Flags SDK offers an open-source library for integrating feature flags in Next.js and SvelteKit applications, supporting a variety of providers through a new adapters architecture. These advancements aim to accelerate iteration, reduce risk, and enable personalized user experiences while maintaining reliable and consistent deployments.

The Vercel AI Accelerator is a six-week program that offers over $2.7M in credits from top AI platforms, including Vercel, AWS, and leading AI startups. Participants will have access to community support, coaching, visibility opportunities, and credits to use various AI services. The program aims to accelerate the development of next-generation AI apps and is open for applications until May 17th.

With v0, you don't have to compromise on performance, accessibility, or SEO best practices that matter for discoverability. Every interface generated is fast, accessible, and SEO-optimized by default, integrating with Next.js and deploying to Vercel, giving you structured metadata, performance tuning, and Server Side Rendering (SSR). v0 generates code that automatically applies Next.js optimizations to improve Core Web Vitals signals, including native image optimization, font optimization without layout shifts, and smart JavaScript bundling and prefetching. This gives your site the technical foundation to support instant page loads, low blocking time, and a consistent visual experience, all key factors for ranking higher in search results. v0 includes built-in support for metadata and structured data using Next.js conventions, with automatic Open Graph and Twitter Card generation, JSON-LD support for rich snippets in search, and SEO-friendly title and meta description management. The platform is also optimized for mobile-first design, responsive by design, and prioritizes accessibility as a ranking signal. When you deploy a v0 project, it runs on a platform optimized for global performance and reliability, with automatic edge caching, real-time Incremental Static Regeneration (ISR), and privacy-friendly analytics. v0 uses the Next.js App Router and Server Components out of the box, with server-rendered by default, offering faster initial load and first paint for users, stronger scores in Google’s page experience metrics, and a clear picture of how Google sees your site without relying on invasive trackers or third-party cookies.

Apple has been restricted by a federal court from continuing to limit the ability of app developers to link to external payment methods, and from charging a 27% fee on those transactions. This ruling gives developers more freedom to offer lower prices for their products on the web, and allows them to build direct relationships with their users. As a result, developers can now create in-app steering that directs users to external payment pages, without the need for scare screens or redirects. The ruling also opens up new opportunities for developers to own the payment experience, pass savings on to users, and improve their checkout flows. This shift marks a significant change for native app developers, who can now choose between Apple's in-app payments and their own checkout experiences. The ruling also has implications for the web, as it makes mobile commerce more viable and allows developers to build fast, secure, and customizable checkout flows on the web.