Hosting cost worksheet
Vercel Pricing for AI Apps
Vercel Hobby is free for personal, non-commercial projects. Pro costs $20 per month, includes one deploying seat and $20 of infrastructure credit, then adds regional usage charges. For AI apps, the bill usually hinges on provisioned memory, transfer, and seats rather than the time spent waiting for a model API.
A generic page-view estimate misses how Vercel now bills Fluid compute. Active CPU pauses while a Function waits on a model or database, but provisioned memory keeps the meter running until the instance winds down — which is most of an AI request's lifetime. So the calculator starts from the usage units Vercel's own dashboard reports, instead of pretending every model request maps to one isolated serverless instance.
To be clear about what this is: a billing model built from published rates, not a measured hosting benchmark. Use the Netlify vs Vercel comparison for the platform decision and the deployment guide once you have picked a host.
Estimate a Vercel Pro bill
| Active CPU | $1.28 |
| Provisioned memory | $1.06 |
| Function invocations | $0.60 |
| Fast Data Transfer overage | $0 |
| Edge request overage | $0 |
| Infrastructure subtotal | $2.94 |
| Included usage credit | −$2.94 |
| Additional deploying seats | $0 |
Estimated Vercel Pro total
$20.00
$20 platform fee, $0 in extra seats, and $0 in infrastructure usage after the included credit.
Estimate uses Vercel's published Cleveland, Portland, and Washington D.C. starting rates: $0.128 per active CPU hour, $0.0106 per provisioned-memory GB-hour, $0.60 per million function invocations, $0.15 per GB over 1 TB of Fast Data Transfer, and $2 per million Edge Requests over 10 million. It excludes model-provider charges, builds, storage, image optimization, add-ons, taxes, and higher-priced regions.
Vercel plan pricing at a glance
| Plan | Platform fee | Who can use it | Included delivery | Compute |
|---|---|---|---|---|
| Hobby | $0 | Personal, non-commercial | 100 GB transfer; 1M Edge Requests | 4 active CPU hours; 360 memory GB-hours; hard caps |
| Pro | $20/mo | Commercial apps and teams; one deploying seat included | 1 TB transfer; 10M Edge Requests | $20 usage credit, then regional on-demand billing |
| Enterprise | Custom | Custom security, controls, regions, and support | Custom | Custom terms and limits |
The limits that matter for an AI app
Commercial use decides the floor.Hobby is restricted to personal, non-commercial work. A customer-facing product, paid client project, or app built by a paid employee belongs on Pro or Enterprise even if its usage fits inside Hobby's quotas.
Streaming is supported. Vercel Functions can pass model output to the browser as chunks arrive. With Fluid compute, Hobby Functions default to and max out at 300 seconds. Pro defaults to 300 seconds and is documented up to 800 seconds, with an opt-in 1,800-second beta for supported Node.js and Python runtimes announced in June 2026.
I/O changes the cost shape. Calling a model API often leaves the Function waiting. That idle wait does not add active CPU time, but allocated memory continues to accrue. Record both dashboard metrics; multiplying wall-clock duration by CPU price will overstate one line and miss the other.
Seats can outrun compute. The $20 Pro fee includes one deploying seat and free read-only Viewer seats. Each additional Owner or Member seat that can configure and deploy projects costs another $20 per month.
When Netlify prices better
Netlify's current Pro plan also starts at $20 per month, but includes unlimited members and 3,000 credits shared across production deploys, compute, bandwidth, and web requests. That can be easier to budget for a collaboration-heavy team. Vercel gives one deploying seat, separate delivery allocations, a $20 infrastructure credit, and regional metering.
Neither billing model is universally cheaper. Put your own usage into the calculators before comparing the totals, then check framework support, Function duration, preview workflow, and observability. The side-by-side comparison page keeps those non-price constraints beside the estimate.
Vercel pricing FAQ
How much does Vercel Pro cost?
Vercel Pro has a $20 monthly platform fee. It includes one deploying seat and $20 in monthly infrastructure credit. Additional deploying Owner or Member seats are $20 each per month; Viewer seats are free. Usage beyond included allocations and credit is billed on demand.
Can a commercial AI app use Vercel Hobby?
No. Vercel's fair-use rules restrict Hobby to personal, non-commercial projects. Commercial deployments require Pro or Enterprise even when their traffic stays below Hobby quotas. Treat the plan rule as the first decision, then use the calculator to estimate infrastructure overage on Pro.
Does Vercel charge while an AI model is generating?
Active CPU billing pauses while a Function waits on an external model API or database. Provisioned memory billing continues while the instance is alive, including I/O waits. The final charge therefore depends on both active CPU hours and memory GB-hours, not only request duration.
Are model API charges included in Vercel pricing?
No. Vercel bills the application infrastructure that runs your code and delivers responses. OpenAI, Anthropic, Google, or another model provider bills tokens separately. Use the AI cost calculator for model usage, then add that result to this hosting estimate.
Official sources checked
- Vercel pricing — plan fees, delivery allocations, and starting overage rates.
- Vercel Pro plan — included seat, usage credit, extra seats, and spend behavior.
- Fluid compute pricing — active CPU, provisioned memory, invocation accounting, and regional rates.
- Vercel Function limits — memory, duration, concurrency, and streaming limits.
- Vercel fair-use guidelines — Hobby commercial-use restriction.