A New Foundation for Vercel Functions

Vercel has spent the past year rebuilding its compute layer on top of its Managed Infrastructure. The first major result is a substantial update to Vercel Functions, focusing on concurrency, developer experience, and runtime performance.

The headline change is a significant increase in automatic concurrency scaling. Pro accounts can now run up to 30,000 functions simultaneously, while Enterprise plans can scale to 100,000, with options for extended concurrency beyond that. These limits are designed to handle erratic traffic spikes common in news and ecommerce. The new scaling behavior is the default for Hobby and Pro tiers and requires no code changes.

Adopting a Web Standard Interface

The original Vercel Functions API relied on the Node.js request and response objects, an Express-like pattern that also shaped the Next.js Pages Router. With Node.js 18 maturing its support for Web platform standards, Vercel Functions now accept the standard Request and produce a Response.

export async function GET(request: Request) {
  const res = await fetch('https://api.vercel.app/blog', { ... } )
  const data = await res.json()
  return Response.json({ data })
}
export async function POST(request: Request) {
  // Can also define other HTTP methods like PUT, PATCH, DELETE, HEAD, and OPTIONS
}

This unified, isomorphic signature mirrors Next.js Route Handlers. You can read cookies, headers, and request bodies using familiar Web APIs, and the standardized interface means the knowledge and examples from MDN or LLM-based coding tools apply directly. The legacy Serverless and Edge Function signatures remain fully supported. To adopt the new style, confirm you are on Node.js 18 (the default) and the latest Vercel CLI.

Streaming Without Configuration

HTTP response streaming, launched for Vercel Functions earlier this year, now works across both Node.js and Edge runtimes with zero configuration, including with the new Web API signature. Streaming enables use cases like LLM-powered chat interfaces and improving initial page load times by deferring slower content past the first paint.

Longer Durations and Better Controls

Pro customers can now run functions for up to five minutes, extending the scope of what serverless workloads are practical. Vercel has also tightened the default duration limits to prevent accidental overuse and introduced spend management controls that send SMS alerts or trigger webhooks when usage crosses a set threshold. Scheduled jobs are covered by Vercel Cron Jobs, which are now generally available and invoke functions on a schedule.

Cold Start Optimizations

Cold boot times have improved across all frameworks through Managed Infrastructure changes, with additional gains specific to Next.js. The App Router bundles server code for APIs and dynamic pages together; Vercel refined the heuristics for bundling external dependencies and trimmed the overall server runtime size. Comparing Next.js 13 against the latest Next.js 14, cold boots are now twice as fast with smaller function bundles.

Diagnosing slow functions is also easier. Runtime Logs now expose detailed metrics for outgoing requests, showing how long individual fetch calls take—often the real bottleneck behind a slow response. Next.js App Router users need no configuration to benefit; Pages Router users should enable the experimental.bundlePagesExternal flag.

Regional Failover for Dynamic Traffic

Vercel's Edge Network has long rerouted static assets around regional outages. That resilience now extends to Node.js runtime functions, which can automatically fail over to another region during an outage. The failover behavior also works with Vercel Secure Compute, maintaining private connections to databases and other internal infrastructure across regions.

Expanded Logging for Debugging

Serverless Functions now support larger log payloads per invocation. Previously capped at 4KB, logs could truncate exactly when a large stack trace needed full inspection. The new limits make it easier to trace faults and correlate errors with customer issues in Runtime Logs.

These changes represent the first iteration of a broader compute overhaul. Vercel says further improvements are planned around latency, scalability, and cost efficiency.