Vercel Cost in 2026: What You'll Actually Pay (with Calculator)

Vercel cost is the question every Next.js team eventually has to answer, and the official pricing page is not going to give you a number you can trust. The Hobby plan is free, the Pro plan starts at $20 per seat per month, and after that you are billed on at least seven separate axes that most "Vercel pricing" articles flatten into a single line called "bandwidth".

This post breaks down what you will actually pay, verified against our Vercel pricing calculator and modelled across four real workloads.

Quick answer: Vercel costs $0 on Hobby (personal projects only), $20 per developer seat per month on Pro with a matching $20 of usage credit per seat, and custom pricing on Enterprise. A real Pro account with three seats and a growing SaaS workload (250k monthly visitors, 5M API requests) lands around $305 per month in our calculator. The number that bites is not the seat cost. It is Fast Data Transfer and Edge Requests above the included tiers.

Verified against Vercel's published rates as of May 2026. All numbers in this post come from our pricing calculator's source-of-truth provider file, not from rough estimates.

Vercel's 2026 Pricing Model in 60 Seconds

Vercel has three plans and roughly nine billable usage axes. The plans:

The Pro plan is where 95% of paying customers live, and it works like this: each developer seat costs $20 per month and includes $20 of usage credit. The included allocations (1 TB Fast Data Transfer and 10M Edge Requests) are free and do not consume credit. Everything else gets charged against the credit first, then billed on demand.

Here are the meters Vercel runs on Pro, with verified May 2026 rates:

The Hobby plan ships hard caps instead of overage billing: 100 GB bandwidth, 1M function invocations, 4 hours of Active CPU, 360 GB-hours of Provisioned Memory. Hit any of them and your project stops serving until next month, or you upgrade to Pro.

The 7 Hidden Costs Most Vercel Comparisons Miss

Most blog posts about Vercel pricing talk about seats and bandwidth and call it a day. Here are the meters that actually move bills, in roughly the order they bite:

• Edge Requests: Every asset on every page view counts. A page with 35 images, scripts, and stylesheets plus a single API call is 36 edge requests. At 2 million page views per month that is 72M edge requests, well over the 10M included tier.
• Active CPU: Vercel's Fluid Compute charges per CPU-second of actual execution. I/O wait is free, which is a real win for serverless Next.js, but it does add a meter most comparisons skip entirely.
• Provisioned Memory: Charged per GB-hour while functions are warm. A 1 GB Next.js route running 250ms per request, called 5M times, generates ~347 GB-hours per month.
• Image Optimization: Vercel bills per source image processed (rates change, so check the current pricing page). A blog with 100 posts and 5 hero variants per device size can quietly hit thousands of source images on first deploy. This is one of the two stealth charges that catch image-heavy sites off guard, and it's deliberately excluded from our calculator because Vercel reprices it more often than the other meters.
• Edge Middleware Invocations: If your middleware.ts runs on every route, including assets, the request count explodes. Auth gates on every request multiply this further.
• Fast Origin Transfer: Data moving from your function back to the edge. Charged at $0.06/GB. Small until you start streaming large JSON or serving uncached responses.
• Vercel Marketplace add-ons: Postgres, KV, Blob, and the AI Gateway are billed separately from your Pro plan. They use the same credit, but the consumption can be heavy: a $0.40/GB Postgres charge plus AI tokens at OpenAI list prices stacks up fast.

The Pro plan's $20 monthly credit absorbs the small stuff. The bills that surprise teams come from one or two of these meters running 10x over the credit because of a single deploy that changed the workload.

What Does Vercel Actually Cost? Four Real Workloads

We built a Vercel pricing calculator that models every meter Vercel bills on. Here are four workload presets we ship in the calculator, with the resulting monthly Vercel bill. You can change any number and watch the bill move.

Workload 1: Hobby (5k monthly visitors, free plan)

A side project or portfolio. 5k visitors, 3 page views each, a single seat, 15 deploys per month.

This is the "Vercel is free" case. It is also the only case where Vercel is free, because the moment you charge a customer, the terms of service require you to move to Pro.

Workload 2: Startup (50k monthly visitors, 3 seats on Pro)

An early-stage SaaS. 50k visitors, 5 page views each, 8 API requests per visitor, 40 deploys per month, 3 deploying seats.

Read this row carefully. The Startup workload generates about $22 of metered usage, all of which is absorbed by the $60 total credit (3 seats × $20). On-demand billing is zero. You are paying for seats, not for usage. This is the cleanest illustration of how Vercel's Pro plan actually works, and it is the case where Vercel feels cheap and predictable.

Workload 3: Growing (250k monthly visitors, 3 seats on Pro)

A profitable B2B SaaS in the $10–50k MRR range. 250k visitors, 8 page views each, 20 API requests per visitor, 100 deploys per month, 3 deploying seats.

The inflection point. You exceeded the included 1 TB bandwidth (overage: 526 GB) and blew past 10M edge requests by 6.5x. The $60 of seat credit gets eaten in days, and on-demand billing kicks in. Edge Requests alone is $130, more than the entire seat cost.

Workload 4: Scale (1.5M monthly visitors, 5 seats on Pro)

A scale-up. 1.5M visitors, 10 page views each, 30 API requests per visitor, 200 deploys per month, 5 deploying seats.

This is where teams start booking the "talk to Vercel sales" meeting. The bill is dominated by two lines: bandwidth at $1,352 and edge requests at $1,270. Seat cost ($100) and Active CPU ($192) are small change in comparison. At this scale, Enterprise pricing or a CDN offload usually beats the published rates, which is why our calculator flags it with a warning.

The takeaway across all four: Vercel's bill scales with bandwidth and edge requests, not seats. Doubling your team adds $20/seat. Doubling your traffic can add hundreds. The inflection point sits somewhere between the Startup and Growing presets, around 1 TB of egress per month.

When Vercel Gets Expensive (and When It Doesn't)

Vercel is expensive when your workload is bandwidth-heavy or middleware-heavy. It is cheap when your workload is mostly static and your team has fewer than 5 paying seats. The fork is structural, not negotiable.

Vercel gets expensive fast for:

• Image-heavy marketing sites that rely on Vercel's Image Optimization and serve large hero images on every page.
• Apps with edge middleware on every route, especially auth checks that run on asset requests too.
• High-traffic SSR routes where Active CPU dominates because every render does real work, not just I/O.
• Anything past 1 TB of monthly egress, where the $0.15/GB rate stacks up and you cannot offload to a free CDN without leaving Vercel's routing.
• Multi-region or video-streaming workloads, where Fast Data Transfer becomes the entire bill.

Vercel stays cheap for:

• Mostly-static sites and docs (under 100 GB egress, mostly SSG).
• Indie SaaS with 1–3 seats and traffic under ~50k visitors per month, where the credit absorbs everything.
• Apps with aggressive caching and minimal edge middleware.
• Teams that value zero-config Next.js deploys, preview environments, and Skew Protection more than the extra $50–150 per month they cost.

The rule of thumb I give MakerKit customers: if your monthly bandwidth is under 1 TB and you do not run edge middleware on every request, your Pro bill is basically the seat count times $20. If either of those changes, switch the calculator open before your next big deploy.

Vercel vs Cheaper Alternatives

We run the pricing calculator hub so you can compare the same workload across six providers side by side. Here is the Growing preset (250k visitors, 5M API requests, 3 seats) on each:

Three things to notice. First, Cloudflare Workers is an order of magnitude cheaper at this workload because they do not charge for egress at all, which is the line that dominates Vercel's bill. Second, a Hetzner-class VPS behind a free Cloudflare CDN runs $30 for the same workload, which is roughly 10x cheaper than Vercel. Third, Render's fixed-instance model gets expensive at this scale because you need an $85 Pro instance plus ~$450 in bandwidth overage (15 × $30 per 100 GB above the 100 GB included tier), before any database.

The honest read: Vercel is rarely the cheapest, but it is rarely the most expensive either. The premium buys you zero-config Next.js, preview deploys, and Skew Protection. Whether that premium is worth $250 per month versus Cloudflare Workers depends on how much engineering time you would burn rebuilding the DX yourself.

We deploy MakerKit to Vercel by default and ship an OpenNext adapter for Cloudflare Workers, a Docker setup for VPS, and standard Node builds for Railway and Render. The kit is portable, so the choice is yours to make on cost.

Is Vercel Worth It? Our Take After Shipping on All Six

This is the section where most "Vercel pricing" articles get squishy. We are not going to. After running MakerKit kits in production on Vercel, Cloudflare Workers (via OpenNext), Railway, Render, and a Hetzner VPS, here is how we actually pick:

Use Vercel when:

• You are pre-PMF or shipping fast, and engineering hours are worth more than the bill.
• You are a Next.js shop and want every new App Router or Turbopack feature on day one.
• You have 1–10 seats and a workload that fits inside the credit.
• Preview deploys and Skew Protection are load-bearing for your team's velocity.

Skip Vercel for Cloudflare Workers when:

• Bandwidth dominates your cost (over 1 TB/month).
• You can live with the OpenNext adapter's edge runtime constraints.
• You want global low latency without paying for it per request.

Skip Vercel for a VPS when:

• You have someone on the team who is comfortable with a Dockerfile and a systemctl command.
• Your traffic spikes are unpredictable but bounded by hardware (CPU, RAM), not by per-request meters.
• You are running long-lived processes (WebSockets, queues, cron) that fight serverless models.
• Cost predictability matters more than autoscaling.

Skip Vercel for Railway when:

• You want containers without operating them, and your team prefers a fixed compute size over per-invocation billing.
• You are running services besides Next.js (Python, Go, background workers) that need to live next to your app.

Skip Vercel for Render when:

• You want predictable fixed-tier pricing, no per-request math, and you accept the bandwidth overage ceiling.
• You need managed Postgres on the same provider without leaving for Supabase or Neon.

Our default recommendation for a new MakerKit kit is still Vercel for the first 6 to 12 months. After that, the right move depends on what dominates your bill. If you are seeing more than $300/month and most of it is bandwidth, the next stop is the Cloudflare Workers calculator or the VPS calculator. If it is compute, Railway or a VPS make sense. We have shipped to all six in production, and the only wrong answer is staying on the wrong tier out of inertia.

For a deeper provider-by-provider breakdown, see our companion piece on Next.js hosting providers compared.

Calculator: Estimate Your Real Vercel Bill

This is what we built and why we wrote this post: every other "Vercel pricing" article on the SERP explains numbers. Ours models them.

Plug your own traffic, API requests, function duration, and team size into the Vercel pricing calculator. Then switch to the calculator hub to compare the same workload against Netlify, Cloudflare Workers, Railway, Render, and a self-hosted VPS, side by side. Every number in this post traces back to that calculator's source code, so you can verify the math against Vercel's published rates.

If you are considering leaving Vercel, we wrote a piece on running Next.js with version skew protection without Vercel that covers the deploy-safety side, which is the thing most teams forget when they migrate off.

Quick Recommendation

Vercel is worth it for:

• Next.js teams that value zero-config deploys and preview environments.
• Workloads under 1 TB bandwidth and minimal edge middleware.
• 1–10 seats, where the credit absorbs most usage.

Skip Vercel if:

• Bandwidth dominates your cost (over 1 TB/month) and Cloudflare Workers' free egress is a better fit.
• You have ops capacity for a VPS and want 10x lower hosting cost.
• You need long-lived processes that fight the serverless model.

Our pick: Vercel for the first year of a Next.js SaaS, then re-evaluate against Cloudflare Workers or a VPS once monthly bandwidth crosses 1 TB. The right call depends on which meter dominates your bill, which is exactly what our calculator is designed to show.

Disclosure: MakerKit ships Next.js SaaS kits that deploy to Vercel by default and include adapters for Cloudflare Workers, VPS, Railway, and Render. We make money when you buy a kit. We do not get paid by Vercel or anyone else on this list.

Next Steps

For a side-by-side breakdown of every Next.js hosting option, including managed providers, edge runtimes, and self-hosted setups, see our guide to the best Next.js hosting providers. If you are choosing the rest of your stack at the same time, read our take on MakerKit's modern SaaS stack for 2026, which covers payments, database, auth, and deployment together.