Vercel Limits for Vibe Coders: Timeouts, Hobby Rules and Spend Caps
By Matt Cullerton, founder of Mavric and Apso
A TikTok from @alanonai (Alan Knox, who runs the "Engineering for Vibe Coders" series) says Vercel is a good place to start an AI-built app but warns about tradeoffs for "long-running background tasks." The advice to plan for scale is reasonable. The specific tradeoff it names was largely addressed by Vercel before the video was posted on 2026-08-25, and the limits that matter more for vibe coders go unmentioned.
How long can a Vercel function run?
Up to 1,800 seconds (30 minutes) on Pro and Enterprise. Vercel's changelog dated 2026-06-15 raised the ceiling from 800 seconds to 1,800 seconds on Node.js and Python. Durations above 800 seconds are in beta and need maxDuration: 1800 in the route config or vercel.json.
The current functions limits page shows:
- Hobby: "300s default and maximum."
- Pro and Enterprise: "300s default, 800s maximum, and 1800s extended maximum Beta."
For work longer than any cap, the same page points to Vercel Workflows for "workloads that require unlimited execution time," which can "pause, resume, and maintain state for minutes to months without duration limits."
Is the Vercel 10-second timeout still real?
Only for old projects. Vercel's limits page applies the 10-second figure to projects "deployed to Vercel before April 23rd 2025 and not using Fluid compute." For those projects the defaults and maximums are Hobby 10s and 60s, Pro 15s and 300s, and Enterprise 15s and 900s.
Projects created after that date run on Fluid compute by default and start at 300 seconds on every plan. A vibe-coded app deployed in 2026 gets a five-minute ceiling on the free tier, thirty times the legacy Hobby default.
Some limits are the same on every plan. Builds cap at 45 minutes per deployment, proxied requests time out at 120 seconds, and each project can have 100 cron jobs. Hobby allows 100 deployments a day and 1 concurrent deployment, against 6,000 a day and up to 500 concurrent builds on Pro.
Can Vercel handle background jobs and WebSockets?
Yes, with features that shipped before the video:
- Vercel Queues entered public beta on 2026-02-27, with durable topics, at-least-once delivery and automatic retries, starting at $0.60 per 1M operations. The docs say "Vercel Queues (Beta) are available on all plans" and list "Defer expensive work" and "Absorb traffic spikes" as uses.
- WebSockets entered public beta on 2026-06-22. Vercel Functions serve WebSocket connections directly, Socket.IO is supported, and connections follow the same limits and pricing as other function invocations on Fluid compute.
Moving some work off Vercel can still make sense. The reason the video gives was two to six months out of date when it posted.
Can you use the Vercel Hobby plan for a business?
No. The video says that once deployed, "Vercel is the one that is providing the web application to your customers." The Hobby plan doc says: "the Hobby plan restricts users to non-commercial, personal use only." An app with paying customers belongs on Pro under Vercel's own terms.
Going over a Hobby quota does not produce a bill. The doc says "if you exceed your usage limits on the Hobby plan, you will have to wait until 30 days have passed before you can use the feature again." Web Analytics resumes after 7 days.
What do Vercel Hobby and Pro cost?
The pricing page lists Hobby at $0 a month and Pro at $20 a month. Pro seats cost "$20 per user / month, while Viewer seats are free."
Hobby includes, per month:
- 100 GB Fast Data Transfer and 10 GB Fast Origin Transfer
- 1,000,000 CDN requests and 1,000,000 function invocations
- 4 CPU-hours of Active CPU and 360 GB-hours of provisioned memory
- 5,000 image transformations and 200 projects
Pro raises Fast Data Transfer to 1 TB and CDN requests to 10M per month.
Active CPU pricing for Fluid compute, announced 2025-06-25, is $0.128 per Active CPU hour and $0.0106 per GB-hour of provisioned memory. The functions limits page notes that "Waiting for I/O (e.g. calling AI models, database queries) does not count towards active CPU time," so a function blocked on an LLM call bills CPU for only part of that wait.
How to stop a surprise Vercel bill
Use Spend Management, which is "available on Enterprise and Pro plans (Enterprise teams require the Flexible Commitment plan)." Hobby lists it as "N/A" because Hobby pauses instead of billing.
Two notes in the doc matter. First, "Setting a spend amount does not stop usage on its own. To stop incurring costs above your spend amount, turn on Pause Production Deployments." Second, the check "happens every few minutes," so "projects can keep serving traffic and accruing usage for several minutes after you cross the spend amount."
Two changelog entries describe the default differently. One from 2024-06-27 says Spend Management pauses production deployments by default. One from 2025-09-09 says deployments "will continue without interruption unless a hard limit is manually configured." The doc page, updated 2026-09-18, is the one to follow, and it says to check the pause toggle yourself.
Bottom line
Long-running work is less of a Vercel problem than the video suggests: functions run up to 30 minutes on Pro, Queues and WebSockets are in beta, and Workflows covers longer jobs. The limits that catch vibe coders are Hobby's non-commercial rule, its 30-day lockout, and spend caps that only stop usage if you turn on pausing.
Sources
- Vercel Pricing: Hobby and Pro prices and included usage.
- Platform limits: deployment, build and cron limits, plus the legacy timeout table.
- Functions limits: current duration limits and Active CPU rules.
- Hobby plan doc: included usage, the non-commercial rule and the 30-day lockout.
- Spend Management doc: plan availability and the pause setting.
- Queues docs: the background work feature.
- Workflows docs: durable execution beyond function limits.
- Functions can now run up to 30 minutes: the 2026-06-15 change.
- Queues now in public beta: the 2026-02-27 launch and pricing.
- WebSocket support in public beta: the 2026-06-22 launch.
- Active CPU pricing for Fluid compute: CPU and memory rates.
- Spend Management enabled by default on Pro: the 2025-09-09 default.
- Spend Management pauses production by default: the 2024-06-27 default.
- Alan Knox: the creator's site and the Engineering for Vibe Coders series.
Published August 25, 2026. Based on a TikTok by @alanonai. Full video notes (markdown)