What causes "Vercel FUNCTION_INVOCATION_TIMEOUT (504) Error"?

Slow database or third-party API calls Code path that never sends a response (missing return, unresolved promise) Function region (default iad1) far from the database or API Long jobs that do not fit in a request Old projects without Fluid compute still on lower limits

How do I fix "Vercel FUNCTION_INVOCATION_TIMEOUT (504) Error"?

1. Make every path return a response: Check Logs for the timed-out request Wrap handlers in try/catch and always return a Response Add timeouts to fetch and database calls 2. Move the function region next to your data: Settings > Functions > Function Region Choose the region closest to your database or main API Redeploy 3. Raise maxDuration (Pro/Enterprise): Enable Fluid compute in Settings > Functions if it is off In a Next.js route add: export const maxDuration = 800 Or set functions.maxDuration in vercel.json Hobby is capped at 300 seconds 4. Move long work out of the request: Return a response immediately and process in the background Use Vercel Workflows or a queue for jobs longer than a few minutes Stream responses for long AI generations

Try or Bye

Browse all analyzed products with real user feedback patterns.

Vercel FUNCTION_INVOCATION_TIMEOUT (504) Error

Ongoing Web (Vercel Functions) Updated October 21, 2026

Quick answer

A 504 FUNCTION_INVOCATION_TIMEOUT means your function ran past its max duration (300s default with Fluid compute). Make sure every code path returns a response, move the function region next to your database or API, and raise maxDuration on Pro (up to 800s).

Most likely fix: Make every path return a response.

Symptoms

Requests hang and return '504: GATEWAY_TIMEOUT' with code FUNCTION_INVOCATION_TIMEOUT, often only in production or under load.

Why it happens

Slow database or third-party API calls
Code path that never sends a response (missing return, unresolved promise)
Function region (default iad1) far from the database or API
Long jobs that do not fit in a request
Old projects without Fluid compute still on lower limits

How to fix it

Try these in order — the most common fix is first.

1. Make every path return a response
1. Check Logs for the timed-out request
2. Wrap handlers in try/catch and always return a Response
3. Add timeouts to fetch and database calls
2. Move the function region next to your data
1. Settings > Functions > Function Region
2. Choose the region closest to your database or main API
3. Redeploy
3. Raise maxDuration (Pro/Enterprise)
1. Enable Fluid compute in Settings > Functions if it is off
2. In a Next.js route add: export const maxDuration = 800
3. Or set functions.maxDuration in vercel.json
4. Hobby is capped at 300 seconds
4. Move long work out of the request
1. Return a response immediately and process in the background
2. Use Vercel Workflows or a queue for jobs longer than a few minutes
3. Stream responses for long AI generations

This guide also covers

vercel function timeout
function_invocation_timeout vercel
vercel 504 gateway timeout
vercel serverless function timeout
vercel maxduration

Still not working?

If Vercel keeps failing, you're not alone. See every Vercel problem users report or the alternatives people switch to.

Other Vercel errors