Ce que Vercel a publié en avril 2026

Vercel adds Grok 4.3 to AI Gateway with 1M token context window

Vercel integrated Grok 4.3 into its AI Gateway, offering a 1M token context window and December 2025 knowledge cutoff. The model improves accuracy, tool calling, and instruction following, accessible …

Stripe introduces Shared Payment Tokens for agent-seller payment delegation

Stripe launched Shared Payment Tokens (SPTs), enabling agents to grant sellers scoped access to customer payment methods for transactions in the US. Sellers receive a token with usage limits (currency…

Pricing change detected for Vercel

Pricing updated for Vercel: - Pro: allowance changed from 1 Developer seat included to 1 developer seat included - Enterprise: promotion changed — Contact sales tea

Vercel Sandbox adds custom tags for environment and team management

Vercel introduced custom tags for Vercel Sandbox in beta, enabling teams to organize, filter, and manage sandboxes by environment, team, or customer. Tags support up to five per sandbox and can be upd…

Vercel enables Pro plan management in Stripe Projects via CLI

Vercel now allows users to provision or upgrade to a Pro plan directly from Stripe Projects using shared payment tokens and the Stripe CLI. This streamlines billing and infrastructure setup by elimina…

Vercel introduces Native Deployment Checks for linting and typechecking

Vercel now supports running lint and typecheck scripts on every deployment in parallel with builds. Native Deployment Checks integrate with package.json scripts, can be marked as required to block pro…

Vercel caps Hobby plan deployment retention at 30 days

Vercel will enforce a 30-day maximum retention policy for Hobby plan deployments starting April 29, 2026, automatically removing older deployments except the 10 most recent production and aliased depl…

Vercel AI Gateway adds Bring Your Own Key and unified model access

Vercel launched AI Gateway, a unified API enabling access to hundreds of AI models from multiple providers via a single endpoint. It supports budget controls, usage monitoring, fallbacks, and embeddin…

Vercel introduces configurable deployment retention policies

Vercel now allows users to configure automatic deletion policies for different deployment states (canceled, errored, preview, production) to enhance security, compliance, and storage efficiency. All p…

Vercel Flags SDK expands framework support and TypeScript features

Vercel released a free, open-source Flags SDK for Next.js and SvelteKit, enabling framework-native feature flagging with TypeScript support, provider-agnostic adapters, and precompute optimization for…

The AI Gateway For Developers

Vercel introduced AI Gateway, a unified interface for accessing hundreds of AI models from multiple providers via a single API key. It offers unified billing, built-in failovers for uptime, and no pla…

Fireside Chat: Inside FLORA’s AI Image Agent Architecture

FLORA built a creative AI agent (FAUNA) that orchestrates 50+ image models and long-running workflows using Vercel Workflows. The integration enabled durable execution, stateful agents, and reduced in…

Vercel adds GPT-5.5 and GPT-5.5 Pro to AI Gateway

Vercel integrated OpenAI’s GPT-5.5 and GPT-5.5 Pro models into its AI Gateway, offering two variants optimized for long-running agentic tasks and high-quality multi-step workflows. The models improve …

Vercel adds DeepSeek V4 Pro and Flash to AI Gateway

Vercel integrated DeepSeek V4 Pro and V4 Flash into its AI Gateway, offering 1M token context windows and specialized capabilities like agentic coding and long-form document generation. Both models ar…

AWS Summit London

AWS Summit London is a one-day, complimentary event scheduled for April 22, 2026, at Excel London, focused on cloud and AI innovation. The event offers over 200 sessions, keynotes, and hands-on experi…

Pricing updated for Vercel: - Pro: allowance changed from 1 Developer seat to 1 Developer seat included - Enterprise: promotion changed — Contact sales for custom p

GPT Image 2 now available on Vercel AI Gateway

Vercel added OpenAI’s GPT Image 2 to its AI Gateway, enabling high-resolution image generation with detailed instruction following, accurate object placement, and support for non-English text. The mod…

Vercel integrates Moonshot AI’s Kimi K2.6 on AI Gateway

Vercel added Moonshot AI’s Kimi K2.6 model to its AI Gateway, enabling developers to deploy it via the unified API. K2.6 excels at long-horizon coding tasks across Rust, Go, and Python, including fron…

Vercel April 2026 security incident

Vercel disclosed a security incident in April 2026 involving unauthorized access to internal systems via a compromised third-party AI tool’s Google Workspace OAuth app. A limited subset of customers w…

Vercel expands framework support with detailed infrastructure matrix and Build Output API v3

Vercel expanded its framework support matrix, now offering first-class integration with popular frontend, backend, and full-stack frameworks like Next.js, SvelteKit, Nuxt, and Remix. The update introd…

Knowledge Base

Vercel expanded its knowledge base with new guides focused on AI-powered application development, including tutorials on building agents with OpenAI Agents SDK, Vercel Sandbox, and Claude Managed Agen…

Vercel launches unified feature flags platform with observability tools

Vercel introduced a unified feature flags platform allowing teams to control feature visibility, run experiments, and deploy safely without redeploying. The platform supports gradual rollouts, A/B tes…

Vercel rebrands as the AI Cloud with new AI-powered tools

Vercel rebranded itself as the AI Cloud, positioning its platform as a unified solution for building, deploying, and scaling AI-powered applications. The documentation highlights new AI-focused tools …

Vercel launches microfrontends support with pricing and documentation

Vercel introduced official documentation and support for microfrontends, enabling teams to split applications into independently deployable units while maintaining a cohesive user experience. The feat…

Vercel expands integration ecosystem with native and connectable accounts

Vercel expanded its integration ecosystem to allow users to connect third-party platforms for AI, databases, headless content, commerce, and more. The platform now offers two types of integrations: na…

Vercel expands Git integration with automatic deployments and enhanced security controls

Vercel now supports automatic deployments for every branch push and merges to the production branch across GitHub, GitLab, Bitbucket, and Azure DevOps Pipelines. This enables preview deployments for e…

Vercel rebrands as 'the AI Cloud' with new AI-first tools and infrastructure

Vercel rebranded itself as 'the AI Cloud,' positioning its platform as a unified solution for building, deploying, and scaling AI-powered applications. The documentation highlights new AI-focused tool…

Vercel launches unified storage suite with Blob, Edge Config, and Marketplace integrations

Vercel introduced a unified storage suite with three products: Vercel Blob for large file storage, Vercel Edge Config for ultra-low-latency global data, and Vercel Marketplace for third-party database…

microsoft-copilot - Overview – Vercel

Vercel has integrated Microsoft Copilot into its platform, enabling AI-driven workflows directly within the Vercel sidebar. The integration allows users to leverage Copilot for tasks such as deploymen…

Zo Computer achieves 20x AI reliability improvement using Vercel's AI Gateway and AI SDK

Zo Computer, a personal AI cloud platform, reduced AI retry rates by 20x (7.5% to 0.34%) and improved chat success to 99.93% after migrating to Vercel's AI Gateway and AI SDK. P99 latency dropped 38% …

Zo Computer cuts AI failures 20x using Vercel’s AI Gateway

Zo Computer achieved a 20x reduction in AI retry rates (7.5% to 0.34%) and improved chat success to 99.93% after migrating to Vercel’s AI SDK and AI Gateway. The switch eliminated custom adapter code …

Vercel updates deployment retention to preserve active branch deployments

Vercel updated deployment retention policies to preserve the latest preview deployment for branches with open or unmerged pull requests, even if they exceed the configured retention window. Previously…

Vercel AI Gateway adds observability dashboards for AI model usage and performance

Vercel introduced AI Gateway observability features to monitor and debug AI model requests. The update provides team-level and project-level dashboards tracking spend, model usage, time-to-first-token…

Vercel adds OpenFeature provider for Vercel Flags

Vercel introduced an OpenFeature provider for Vercel Flags, enabling developers to use the vendor-neutral OpenFeature API while Vercel manages feature flags. The integration supports typed evaluation …

Vercel launches Custom Reporting for AI Gateway with granular cost and usage tracking

Vercel introduced Custom Reporting for its AI Gateway, enabling detailed cost and token tracking by model, user, tag, provider, or credential type. The feature is in beta and available on Pro and Ente…

Claude Opus 4.7

Vercel introduced Claude Opus 4.7, enhancing its coding and agentic capabilities with improved performance on complex, multi-step tasks and more reliable agentic execution. The update also strengthens…

microsoft-365 - Overview – Vercel

Vercel and Microsoft announced a partnership to integrate Microsoft 365 workflows into Vercel's deployment and development platform. The integration allows developers to leverage Microsoft 365 tools (…

Vercel Workflows reaches GA with durable execution for agents and backends

Vercel announced the general availability of Vercel Workflows, a programming model for building durable, long-running agents and backends without separate orchestration services. Workflows eliminates …

Vercel adds Claude Opus 4.7 to AI Gateway with agentic workflows and task budgets

Vercel added Anthropic's Claude Opus 4.7 to its AI Gateway, enabling long-running agentic workflows with improved memory, high-resolution image support, and programmatic tool-calling for visual data a…

Vercel Flags reaches general availability with native SDKs and OpenFeature support

Vercel announced the general availability of Vercel Flags, a built-in feature flag provider for its platform. The solution allows teams to create, manage, and target feature flags with user segments a…

Vercel Introduces Turbo Build Machines and Production Build Prioritization for Pro Projects

Vercel has introduced Turbo build machines as the default for new Pro projects, replacing standard build machines to accelerate deployment speeds. These Turbo machines offer significantly higher compu…

Seedance 2.0 Video Generation Now Available on Vercel's AI Gateway

Vercel now offers ByteDance’s Seedance 2.0 video generation model on its AI Gateway, eliminating the need for separate provider accounts. The model is available in two variants: Standard for highest q…

Reduced pricing for Turbo build machines

Vercel reduced the pricing of its Turbo build machines by 16%, lowering the cost to $0.0035 per CPU per minute. Turbo machines with 30 CPUs now cost $0.105 per minute, down from $0.126, while Enhanced…

@vercel/sandbox

Vercel introduced @vercel/sandbox, a tool enabling arbitrary code execution in isolated, ephemeral Linux VMs powered by Firecracker MicroVMs. The SDK supports Node.js 22+ and integrates with Vercel’s …

Vercel Launches Observability Plus for Advanced Debugging

Vercel introduced **Observability Plus**, a paid upgrade for Pro and Enterprise teams that unlocks advanced data exploration, longer retention, and granular metrics for debugging issues. The upgrade e…

Vercel Introduces Agent Investigation for AI-Powered Debugging

Vercel has introduced Agent Investigation, an AI-powered feature that automatically analyzes error alerts in logs and metrics to identify root causes, replacing manual debugging. The tool runs investi…

Logs filtering for Vercel Workflows now available

Vercel has introduced log filtering for Workflows, enabling users to view all logs associated with a specific workflow run in a single location. Previously, logs were scattered across individual reque…

Elastic Build Machines is now GA

Vercel has made Elastic Build Machines generally available for all Pro and Enterprise customers, now serving as the default for new Pro teams. The system dynamically assigns build machines tailored to…

Vercel Integrates Anthropic Messages API into AI Gateway

Vercel has integrated the Anthropic Messages API into its AI Gateway, allowing developers to route Anthropic SDK tools like Claude Code through a unified gateway by changing only the base URL to `http…

Claude Opus 4.6

Vercel has integrated Anthropic’s Claude Opus 4.6 model into its AI Gateway, positioning it as the leading model for coding and professional work. The update enables agents to handle complex software …

Django Notes App

Vercel introduced a new Django-based note-taking app template designed to demonstrate server-side rendering, URL routing, forms, and ORM using Django. The template defaults to SQLite for local develop…

Building a community adapter for Chat SDK

Chat SDK introduced a new guide for building community adapters, enabling third-party developers to create and publish adapters for any messaging platform not natively supported by the SDK. The guide …

Vercel Firewall Enhances DDoS Mitigation Across All Plans

Vercel now provides automatic DDoS mitigation for all deployments, regardless of plan, blocking suspicious traffic without additional charges. The system monitors, filters, and scales resources in rea…

Vercel Expands Framework Support Documentation with Feature Support Matrix

Vercel has expanded its official documentation to explicitly list and detail support for additional frameworks beyond its core offerings, including Angular, Astro, Brunch, Django, Docusaurus, Dojo, El…

Vercel Introduces Fluid Compute Pricing for Functions

Vercel introduced a new fluid compute pricing model for its Functions, replacing traditional billing with a granular, resource-based approach. Pricing now varies by region and is calculated separately…

Vercel Launches Interaction Timing Tool for Performance Insights

Vercel introduced the Interaction Timing Tool, a new feature within the Vercel Toolbar that allows developers to inspect interaction latency in detail and receive notifications for interactions exceed…

A Better Way to Build Agents: Workflows

Luma announced the launch of Vercel Workflows, a new product designed to enable developers to build durable, production-ready AI agents in TypeScript. The announcement was made during an exclusive bui…

Vercel Launches Accessibility Audit Tool in Vercel Toolbar

Vercel introduced an Accessibility Audit Tool within its Vercel Toolbar that automatically checks Web Content Accessibility Guidelines (WCAG) 2.0 level A and AA rules, grouping issues by severity (Cri…

Vercel Introduces Custom Environments for Advanced Deployment Workflows

Vercel introduced Custom Environments for its Pro and Enterprise teams, expanding beyond the default Local, Preview, and Production environments. Custom Environments allow teams to create specialized …

Contributor Covenant Code of Conduct

The Contributor Covenant Code of Conduct, version 1.4, establishes a standardized framework for fostering inclusive and harassment-free participation in open-source projects and communities. It outlin…

Anomaly alerts are now generally available

Vercel has made anomaly alerts generally available for teams using Observability Plus, enabling real-time detection of unexpected behavior in applications such as unusual usage patterns or error spike…

Copy-to-Prompt instructions now available for Flags

Vercel has introduced a new feature in its Flags product that allows users to copy flag definitions directly from the instructions pane on the feature flags details page. This update enables teams to …

Anomaly alert configuration now available

Vercel has introduced granular anomaly alert configuration, allowing users to customize which projects, metrics, HTTP status codes, and specific routes trigger alerts for unexpected spikes or errors. …

Vercel Unveils Agentic Infrastructure for AI-Native Software Development

Vercel introduces Agentic Infrastructure, a new paradigm for deploying and managing AI-native software built and operated by autonomous coding agents. The shift is driven by rapid adoption: over 30% o…

Zero-configuration Django support

Vercel now supports Django applications with zero-configuration deployment, eliminating the need for manual redirects in `vercel.json` or reliance on the `/api` folder. This change simplifies the depl…

Pricing updated for Vercel: - Hobby: new promotion — Free forever - Pro: new promotion — $20 included credit to use across resources - Pro: added annual billing at $20/mo (0% off) - Pro: now includes

Vercel Expands AI Gateway with Team-Wide Zero Data Retention Compliance

Vercel has expanded its AI Gateway with Zero Data Retention (ZDR) compliance features, addressing fragmented data policies across AI model providers. The update introduces team-wide ZDR enforcement fo…

Vercel CLI now supports Sandbox management directly

Vercel has integrated Sandbox management directly into its CLI via a new `vercel sandbox` subcommand, eliminating the need for a separate tool. This change streamlines the development workflow by unif…

Summary of CVE-2026-23869

Vercel disclosed a high-severity vulnerability (CVSS 7.5) in React Server Components, specifically affecting Next.js versions 13.x through 16.x when using the App Router. The flaw allows specially cra…

Vercel Sandbox now supports up to 32 vCPU + 64 GB RAM configurations

Vercel has expanded its Sandbox offering for Enterprise customers by introducing configurations with up to 32 virtual CPUs (vCPUs) and 64 GB of RAM. This upgrade enables the execution of large-scale, …

Manage Vercel Microfrontends with AI Agents and the CLI

Vercel has introduced two new tools for managing Microfrontends: an AI skill for coding agents and expanded CLI commands. The AI skill enables natural language prompts to automate the creation of micr…

Opus 4.6 Fast Mode available on AI Gateway

Vercel has introduced Fast Mode for Claude Opus 4.6 on its AI Gateway, delivering 2.5x faster output token speeds while maintaining the same model intelligence. This experimental feature enables new u…

Query and visualize workflow data in Vercel Observability

Vercel has introduced a new query builder feature in Observability Plus, enabling Pro and Enterprise teams to create custom queries on Vercel Workflows data. This tool allows users to visualize key me…

Chat SDK adds Liveblocks support

Vercel’s Chat SDK has introduced official support for Liveblocks, enabling bots to interact with Liveblocks Comments threads through a new Liveblocks adapter. Teams can now build bots capable of posti…

GLM 5.1 on AI Gateway

Vercel has integrated GLM 5.1, a new AI model from Z.ai, into its AI Gateway platform. GLM 5.1 is designed for long-horizon autonomous tasks, enabling continuous operation on single tasks through iter…

Vercel CLI Commands Now Scoped to Local Directory

Vercel has updated its CLI commands to scope local directory operations by default, changing how users interact with projects and domains. Previously, commands like `vc project ls` and `vc domains ls`…

Team-wide Zero Data Retention and prompt training controls now on AI Gateway

Vercel has introduced team-wide Zero Data Retention (ZDR) and prompt training controls for its AI Gateway, eliminating the need for individual provider opt-outs. The new feature automatically routes r…

Query and manage Marketplace databases from the dashboard

Vercel has introduced a new capability within its dashboard to query, manage, and edit Marketplace database integrations directly, eliminating the need for external tools like `psql` or third-party da…

roadmap - Overview – Vercel

Vercel has announced that its AI-powered deployment agent, Vercel Agent, is currently in an early preview phase and is not recommended for production environments due to potential inaccuracies and err…

How to build AI Agents with Vercel and the AI SDK

Vercel has published a guide detailing how to build, deploy, and scale AI agents using its AI SDK and infrastructure. The guide explains the process of creating multi-step AI agents that can call larg…

Vercel AI Gateway Introduces Disallow Prompt Training Feature

Vercel has introduced a new capability in its AI Gateway service that allows users to explicitly disallow the use of their prompt data for model training by AI providers. This feature, available at no…

Vercel Introduces Automated & Agent Access for Deployment Protection

Vercel has introduced two methods to bypass Deployment Protection for automated systems like AI agents, CI/CD pipelines, MCP servers, and end-to-end testing tools that cannot complete browser-based au…

Vercel Optimizes Sandbox Snapshots with Parallelization and Caching

Vercel optimized its Sandbox filesystem snapshots to reduce restore times from over 40 seconds to under one second. The changes involved parallelizing S3 downloads using HTTP Range headers, multi-thre…

Base fee for Observability Plus removed

Vercel has removed the $10 base fee for its Observability Plus product, transitioning to a pay-per-observability-event pricing model. This change eliminates a fixed cost for all customers, making the …

Nuxt MCP Toolkit Launched for AI Feature Development

Vercel has introduced the Nuxt MCP Toolkit, enabling developers to build Model Context Protocol (MCP) servers directly within Nuxt applications. This new module allows teams to define tools with Zod v…

Qwen 3.6 Plus on AI Gateway

Vercel has integrated Alibaba’s Qwen 3.6 Plus model into its AI Gateway, a unified API service for accessing large language models. The new model, an upgrade from Qwen 3.5 Plus, introduces enhanced ag…

Vercel launches AI Gateway for unified AI model access

Vercel introduced AI Gateway, a unified API enabling access to hundreds of AI models from multiple providers through a single endpoint. It includes built-in budget controls, usage monitoring, automati…

Zero-configuration Go backend support

Vercel has introduced zero-configuration deployment for Go API backends, enabling developers to deploy Go servers as first-class backends without manual configuration in `vercel.json` or the `/api` fo…

Gemma 4 Models Now Available on Vercel AI Gateway

Vercel has integrated Google’s Gemma 4 models—specifically the 26B (MoE) and 31B (Dense) variants—into its AI Gateway platform. These models, built on the same architecture as Google’s Gemini 3, suppo…

Custom Class Serialization in Workflow DevKit

Workflow DevKit now supports custom class serialization, enabling developers to pass custom class instances between workflow and step functions. Previously, only standard JavaScript types were support…

Vercel Public Beta Agreement

Vercel has introduced a Public Beta Agreement outlining the terms for users accessing its products during the Public Beta release phase. The agreement, effective upon user acceptance, governs the use …

Search code, repositories, users, issues, pull requests...

Vercel has released the AI SDK, a provider-agnostic TypeScript toolkit designed to simplify the development of AI-powered applications and agents. The SDK supports popular UI frameworks like Next.js, …

Svelte is an open-source web development compiler that converts declarative components into efficient JavaScript for DOM updates, positioning itself as a modern alternative to traditional frameworks.

The GitHub repository for Nuxt, a free and open-source framework for building type-safe, performant, and production-grade full-stack web applications and websites with Vue.js, is presented. Nuxt simpl

The GitHub repository for Nitro, a server toolkit for Vite applications, highlights a transition from v2 to v3, with v3 currently in development and v2 remaining the stable release. Nitro extends Vite

The Workflow Development Kit (SDK) by Vercel enables developers to build durable, reliable, and observable asynchronous JavaScript applications and AI agents. The SDK allows apps to suspend, resume, a

SWR is a React Hooks library for data fetching, leveraging the stale-while-revalidate (SWR) cache invalidation strategy to optimize performance. The library simplifies data fetching logic by returning

Turborepo is a high-performance build system designed for JavaScript and TypeScript codebases, developed using Rust. The repository serves as the primary hub for the project, offering resources for ge

Pricing updated for Vercel: - New tier: Vercel Agent ($0.30/mo) - New tier: Viewer seat (Custom pricing) - New tier: Billing seat (Custom pricing)

GLM 5V Turbo Now Available on Vercel AI Gateway

Vercel has integrated GLM 5V Turbo, a multimodal coding model from Z.ai, into its AI Gateway platform. The model specializes in converting screenshots and designs into code, debugging visually, and au…

How Waldium made a blog platform work for humans and AI alike

Waldium, a two-person YC-backed startup, has developed an agentic CMS platform that automates content research and creation while enabling AI agents to directly query customer blogs via MCP (Model Con…

Vercel Queues now supports 7-day message TTL

Vercel Queues now supports a maximum message Time-To-Live (TTL) and delivery delay of 7 days, an increase from the previous 24-hour limit. The default TTL remains 24 hours with a 0-second delivery del…

Chat SDK adds Zernio support

Vercel’s Chat SDK has introduced official support for Zernio, a unified social media API, via a new Zernio adapter. This integration enables developers to build bots that interact across multiple plat…