eve is an open-source agent framework for building and scaling agents.
▪︎ Durable execution
▪︎ Sandboxed compute
▪︎ Human-in-the-loop
▪︎ Subagents and evals
Available now in public preview ↓
vercel.com/changelog/introdu…
Workflow Dev Kit: reliability-as-code
• "𝚞𝚜𝚎 𝚠𝚘𝚛𝚔𝚏𝚕𝚘𝚠" + "𝚞𝚜𝚎 𝚜𝚝𝚎𝚙"
• Open source and MIT-licensed
• Designed for AI agents and backends
• Backend-agnostic (Vercel or self-hosted)
• Framework-agnostic (Next.js, Nitro & more)
useworkflow.dev
Vercel Agent is now available in public beta.
• Runs AI code reviews for correctness, security, & performance
• Generates and validates patches before they reach your PR
• Reviews TypeScript, Python, Go, and more
Try it now with $100 in free credit.
vercel.com/changelog/ai-code…
Zero-config support for Express, now on @vercel Fluid compute:
• Full @nodejs runtime
• Cold start prevention
• Multi-AZ and multi-region failover
• Efficient, cost-effective concurrency
Deploy Express backends with just 𝚊𝚙𝚙.𝚝𝚜.
vercel.com/changelog/zero-co…
Zero-configuration support for NestJS backends, now on Vercel.
NestJS joins frameworks like Express, FastAPI, Flask, and Hono, all supported via framework-defined infrastructure.
vercel.com/changelog/zero-co…
Vercel Queues offload work to the background, allowing slow or long-running jobs to guarantee completion.
Now available in Limited Beta.
vercel.com/changelog/vercel-…
Save your holiday traffic for customers, not bots.
Enterprise and Pro teams can use Vercel BotID Deep Analysis free of charge through January 15.
vercel.com/changelog/free-bo…
Vercel's MCP server now works with Cursor.
Access projects and deployments, analyze logs, and search docs directly from within Cursor.
vercel.com/changelog/cursor-…
Microfrontends on Vercel are now generally available.
• Break monoliths into independent apps
• Already serving nearly 1B microfrontends requests per day
• In production for over 250 teams, like Cursor, The Weather Company, and A+E Networks
vercel.com/changelog/microfr…
Microfrontends support is now in Public Beta and available to teams on all plans.
Split large apps into smaller, independently deployed ones with their own tech stacks.
Vercel handles shared routing and fast navigation for smooth user experiences across microfrontends.
Vercel has partnered with Z.ai to offer the lowest pricing for GLM 4.6.
• Available via Vercel AI Gateway
• $0.45 / M input tokens
• $1.80 / M output tokens
• Set the model to 𝚣𝚊𝚒/𝚐𝚕𝚖-𝟺.𝟼 to start building
vercel.fyi/glm-4.6
Vercel now supports Nitro applications, a backend toolkit for building web servers, with zero-configuration.
Nitro powers frameworks like Nuxt.js, TanStack Start, and SolidStart.
vercel.com/changelog/zero-co…
The Vercel Sandbox CLI is now available.
▲ ~ / 𝚗𝚙𝚡 𝚜𝚊𝚗𝚍𝚋𝚘𝚡 𝚛𝚞𝚗
Create, run, and manage secure, ephemeral Sandbox environments right from your terminal.
You can now deploy FastHTML projects to Vercel with zero configuration.
FastHTML is a new Python framework for fast, scalable web applications with minimal, compact code. It builds on top of popular foundations like ASGI and HTMX.
vercel.com/changelog/fasthtm…
You can now connect ChatGPT to Vercel MCP.
ChatGPT can now securely access your Vercel projects, deployments, build logs, and more.
vercel.com/changelog/chatgpt…
Stripe is now available in beta on the Vercel Marketplace.
Provision a Stripe claimable sandbox with zero setup to:
• Test ecommerce checkouts
• Validate SaaS subscriptions
• Share demo-ready payment flows
• Onboard devs instantly
vercel.com/changelog/stripe-…
We've rebuilt Grep with Next.js.
• Improved performance with Partial Prerendering (PPR)
• Now indexing 1,000,000+ repositories
• Support for dark mode
vercel.com/changelog/dark-mo…
Vercel TypeScript SDK
• Full type safety for accessing the Vercel REST API with Zod schemas
• New documentation for every function, argument, and type
• Better tree-shaking support with optional standalone functions
vercel.fyi/sdk
We've reduced the price of Web Analytics by up to 80% and increased the number of included events:
• Hobby: 2.5K → 50K events per month
• Pro: 25K → 100K events per month
vercel.com/changelog/up-to-8…
Vercel BotID is now available for every web app.
This extends BotID's invisible bot protection, already available for Next.js, to all JavaScript frameworks including SvelteKit, Nuxt, and more.
vercel.com/changelog/botid-n…
The free tier can now run Vercel Functions 6x longer, up to 60 seconds. This is beneficial for streaming responses, especially when building AI applications.
vercel.com/changelog/vercel-…
You can now access Qwen3-Coder through Vercel AI Gateway, with no provider accounts required.
Set the AI SDK v5 𝚖𝚘𝚍𝚎𝚕 to '𝚊𝚕𝚒𝚋𝚊𝚋𝚊/𝚚𝚠𝚎𝚗𝟹-𝚌𝚘𝚍𝚎𝚛' to start building.
vercel.com/changelog/qwen3-c…
You can now deploy Hono backends with zero-configuration.
Hono apps deployed to Vercel benefit from Fluid compute, with Active CPU pricing, automatic cold start optimizations, background processing, and more.
vercel.com/changelog/deploy-…
Welcome to the Vercel Changelog!
On this account, we'll tweet features and other changes we're making to @vercel. Company updates will continue to be posted on our blog.
Follow us to track how we work constantly to make Vercel better with your feedback.
vercel.com/changelog
Vercel Remote Cache is now free for all plans.
Never do the same work twice in your Turborepo or Nx repository with a shared store of build outputs and logs for your team.
vercel.com/changelog/free-ve…
Go full Turbo.
Turbopack is now the default for all Next.js 16 apps.
Pair it with Vercel Turbo build machines for even faster builds.
vercel.com/changelog/turbo-b…
Vercel Sandbox securely runs untrusted code in isolated cloud environments, like AI-generated code.
Create ephemeral, isolated microVMs using the new Sandbox SDK, with up to 45m execution times.
Now in Beta and available to customers on all plans.
vercel.com/changelog/run-unt…
You can now access @openai's new open models through Vercel AI Gateway.
Set the AI SDK v5 𝚖𝚘𝚍𝚎𝚕 to either 'openai/gpt-oss-20b' or 'openai/gpt-oss-120b' to start building.
Multiple redundant inference providers. No accounts required.
vercel.com/changelog/gpt-oss…
Vercel's MCP server now supports Devin, Raycast, Windsurf, and Goose.
You can now give more AI agents secure access to your projects, deployments, build logs, and more.
vercel.com/changelog/devin-r…
New challenger on the Next.js Evals: Cursor's Composer 1.
4th on the leaderboard and 4× faster than similarly intelligent models.
See the full board ↓ vercel.fyi/next-evals
Connections to all Vercel deployments are now secured with post-quantum encryption to protect against future quantum computing attacks.
All modern browsers will use these new algorithms automatically during TLS handshakes.
vercel.com/changelog/post-qu…
Vercel now natively supports bulk redirects, scaling up to 1 million redirects per project.
Set 𝚋𝚞𝚕𝚔𝚁𝚎𝚍𝚒𝚛𝚎𝚌𝚝𝚜𝙿𝚊𝚝𝚑 to your redirects folder and they'll be automatically imported at build time.
Available for Pro and Enterprise teams.
vercel.com/changelog/bulk-re…
AI Gateway is now in Beta, giving you a single endpoint to access a wide range of AI models across providers, with better uptime, faster responses, no lock-in.
Use models from providers like OpenAI, xAI, Anthropic, Google, and more.
vercel.com/changelog/ai-gate…
Vercel Functions now use bytecode caching by default.
Bytecode caching reduces cold start times by up to 60%, exceeding our initial benchmarks and observations.
vercel.com/changelog/bytecod…
Skew Protection—a novel mechanism to eliminate version skew between web clients and servers.
Now generally available for paid plans. Supported by Next.js, SvelteKit, and more frameworks soon.
vercel.com/changelog/skew-pr…
Introducing the Jira integration for Vercel.
Comment, review, and share feedback with your team on deployments. Then, instantly convert the conversation into a Jira issue.
vercel.com/changelog/convert…
Python builds are up to 65% faster.
Vercel now defaults to uv for Python projects, bringing faster installs and compatibility with more dependency file formats.
vercel.com/changelog/python-…
You can now block a Vercel deployment from going to production until GitHub Actions finish successfully.
Available for all GitHub-connected projects.
vercel.com/changelog/block-v…
The Vercel Python SDK is now available in beta.
𝚙𝚒𝚙 𝚒𝚗𝚜𝚝𝚊𝚕𝚕 𝚟𝚎𝚛𝚌𝚎𝚕
Use Python SDK to run untrusted code in Vercel Sandboxes, interact with Blob storage, control the Runtime Cache API, and more.
vercel.com/changelog/vercel-…
MiniMax M2 is now on Vercel AI Gateway.
• Free until Nov 7 2025
• Set model to 𝚖𝚒𝚗𝚒𝚖𝚊𝚡/𝚖𝚒𝚗𝚒𝚖𝚊𝚡-𝚖𝟸
• Open source model for agentic use
Integrated in collaboration with @minimax_ai's engineering team to boost reliability & performance.
vercel.com/changelog/minimax…
Vercel Blob, our frontend-friendly solution for storing and serving files, is now available in public beta for Hobby and Pro customers.
vercel.com/changelog/vercel-…
Node.js 14 and 16 are being deprecated on August 15, 2023.
While existing deployments with functions will not be affected, we encourage updating to Node.js 18.
vercel.com/changelog/node-js…
We build Vercel on Vercel.
Vercel AI Gateway runs on Fluid compute, on the same network as you:
• Concurrency keeps CPUs hot, costs low
• Billed for CPU only when active
• Anycast routing for low latency
• Redis + in-memory caching
• Built-in o11y
vercel.com/blog/how-ai-gatew…
Aliasing for multi-tenant applications is now up to 95% faster.
Multi-tenant applications on Vercel often point thousands of domains to the same deployment through aliasing. With these optimizations, aliasing now completes in seconds instead of minutes.
vercel.com/changelog/faster-…
Cape Town is now available on the Vercel Edge Network, resulting in 50% faster Time To First Byte (TTFB) in South Africa.
vercel.com/changelog/cape-to…
It's now possible to tailor responses to visitors without performance trade-offs using a new property in `next.config.js` and `vercel.json`.
vercel.com/changelog/routing…
You can now directly export standard fetch web handlers in Vercel Functions.
Works with any fetch-compatible framework, or none at all.
vercel.com/changelog/node-js…
You can now run Vercel Sandboxes for up to 5 hours (up from 45 minutes).
𝚊𝚠𝚊𝚒𝚝 𝚂𝚊𝚗𝚍𝚋𝚘𝚡.𝚌𝚛𝚎𝚊𝚝𝚎({
𝚝𝚒𝚖𝚎𝚘𝚞𝚝: 𝟻 * 𝟼𝟶 * 𝟼𝟶 * 𝟷𝟶𝟶𝟶, // 𝟻 𝚑𝚘𝚞𝚛s
});
This new max duration is available for Pro & Enterprise teams.
vercel.com/changelog/vercel-…
You can now deploy xmcp, a framework for building and shipping MCP applications with TypeScript, to Vercel with zero-configuration.
vercel.com/changelog/deploy-…
Vercel Remote Cache is now generally available on all plans. Store, distribute, and reuse artifacts for faster builds across team members and CI.
vercel.com/changelog/vercel-…