March 26, 2026 (1mo ago)
Written by Temps Team
Last updated March 26, 2026 (1mo ago)
Vercel Pro costs $20 per seat per month, includes 1TB of bandwidth and 10 million edge requests, and bills overages at $0.15/GB according to Vercel's pricing page. The free Hobby plan supports one developer with 100GB bandwidth and 1 million function invocations -- but it's restricted to non-commercial use. After Vercel's September 2025 pricing restructuring introduced credit-based billing and moved several enterprise features down to Pro, most older pricing guides are now outdated.
This guide covers every billable dimension of Vercel's 2026 pricing: plans, compute, bandwidth, build minutes, databases, add-ons, and the costs that are easy to overlook. All figures are sourced from Vercel's pricing page, last updated February 27, 2026.
Related: How to migrate from Vercel to self-hosted
TL;DR: Vercel Pro is $20/seat/month with $0.15/GB bandwidth overage after 1TB, as listed on Vercel's pricing page. A 5-person team pays $100/month before any usage. Build costs alone can hit $378/month with Turbo machines. Free tier is non-commercial only. Eight pricing dimensions worth watching -- including DDoS billing and three-dimensional function charges -- make cost estimation more nuanced than it first appears.
Vercel offers three tiers: a free Hobby plan, Pro at $20/seat/month, and Enterprise starting around $45,000/year based on a median of 63 purchases tracked by Vendr's buyer guide. Each tier includes different resource allowances, and overages apply once you exceed them. The September 2025 restructuring added a $20 flexible spending credit to Pro plans and reshuffled which features require Enterprise.
Here's the current plan comparison:
| Hobby (Free) | Pro ($20/seat/mo) | Enterprise (~$45K/yr) | |
|---|---|---|---|
| Seats | 1 | Unlimited ($20/each) | Custom |
| Viewer Seats | -- | Unlimited (free) | Unlimited (free) |
| Bandwidth | 100GB | 1TB | Custom |
| Edge Requests | 1M | 10M | Custom |
| Function Invocations | 1M | Included in compute | Custom |
| Compute (CPU) | 4 CPU-hours (paused on limit) | $0.128/CPU-hr | Custom |
| Build Minutes | 100 min | Pay-per-use | Custom |
| Spending Credit | -- | $20/month | Custom |
| SLA | None | 99.9% | 99.99% |
| Commercial Use | No | Yes | Yes |
The $20 spending credit offsets usage charges but doesn't cover add-on subscriptions like Speed Insights or SAML SSO. It's essentially a buffer for compute and bandwidth overages.
Vercel's Enterprise tier costs approximately $45,000 per year at the median, according to Vendr's buyer guide based on 63 tracked purchases. Pro plans start at $20/seat/month with a $20 flexible spending credit that covers usage overages but not add-on subscriptions.
Related: Next.js hosting cost calculator — calculate your exact Vercel bill
The Hobby plan gives you 100GB bandwidth, 1 million edge requests, 1 million function invocations, and 4 CPU-hours per month -- all free, as detailed on Vercel's pricing page. When you hit any limit, your project pauses rather than billing you. That sounds safe, but there's a significant restriction: Hobby is non-commercial only. Running a SaaS product, a storefront, or even a monetized blog violates the terms.
Here's what the free tier actually looks like in practice:
| Resource | Limit | Real-World Equivalent |
|---|---|---|
| Bandwidth | 100GB/month | ~40,000 page views (2.5MB avg page) |
| Edge Requests | 1M/month | ~33,000 daily requests |
| Function Invocations | 1M/month | Varies by app architecture |
| Compute | 4 CPU-hours | ~240 min of function execution |
| Build Minutes | 100/month | ~20 builds at 5 min each |
| Projects | Unlimited | But resources are shared |
Four CPU-hours sounds generous until you run server-side rendered pages. A Next.js app with dynamic routes that averages 50ms of CPU per request burns through 4 CPU-hours in about 288,000 requests. That's roughly 9,600 requests per day -- fine for a portfolio site, tight for anything with an API layer.
Build minutes are the other bottleneck. If your Next.js project takes 5 minutes to build, you get 20 deployments per month. Teams doing continuous deployment hit that in a week.
And remember: when limits trigger, your site goes down. There's no overage billing on Hobby -- just a hard pause.
Related: 5 ways to deploy a Next.js app in 2026
Pro starts at $20 per seat per month with a $20 flexible spending credit included, per Vercel's pricing page. Viewer seats are unlimited and free, so non-deploying team members don't add cost. Beyond the base, Vercel charges across five billable dimensions: bandwidth, edge requests, compute (CPU + memory), function invocations, and build minutes.
| Resource | Included | Overage |
|---|---|---|
| Bandwidth | 1TB/month | $0.15/GB |
| Edge Requests | 10M/month | $2/million |
A content-heavy site serving 2TB per month pays $150 in bandwidth overages alone. Edge requests rarely cause bill shock for most apps, but middleware-heavy architectures can change that -- more on that below.
This is where Vercel's billing gets three-dimensional. You're charged for CPU time, provisioned memory, and invocations separately:
| Dimension | Rate |
|---|---|
| Active CPU | $0.128/CPU-hour (iad1 region) |
| Provisioned Memory | $0.0106/GB-hour |
| Function Invocations | $0.60/million |
Why does the region matter? Vercel prices compute by region. Sao Paulo (gru1) costs 73% more than US East (iad1) for the same CPU-hour, as noted on Vercel's pricing page. If your users are in Brazil but you haven't checked regional multipliers, your bill might surprise you.
Build minute pricing depends on which machine tier you use:
| Machine Tier | Specs | Rate |
|---|---|---|
| Standard | 4 vCPU, 8GB RAM | $0.014/min |
| Enhanced | 8 vCPU, 16GB RAM | $0.028/min |
| Turbo | 30 vCPU, 60GB RAM | $0.126/min |
Since February 2026, Turbo machines are the default for new Pro projects, as announced in the Vercel Changelog. That matters because the cost difference is 9x compared to Standard. We'll break down the math in the preview deployments section.
Here's how to estimate your Pro bill:
Monthly cost = (seats x $20)
- $20 credit
+ max(0, bandwidth_GB - 1000) x $0.15
+ max(0, edge_requests - 10M) x $0.000002
+ cpu_hours x $0.128
+ memory_GB_hours x $0.0106
+ (invocations / 1M) x $0.60
+ build_minutes x rate_per_tier
+ add-on subscriptions
Most pricing calculators ignore that compute is three-dimensional (CPU + memory + invocations) and that regional multipliers apply. A function running in Sao Paulo with 1GB memory costs roughly 2.4x what the same function costs in US East when you stack the multipliers.
Vercel Pro compute billing operates across three dimensions: $0.128/CPU-hour, $0.0106/GB-hour for memory, and $0.60/million invocations per Vercel's pricing page. Regional pricing varies significantly -- Sao Paulo costs 73% more than US East for CPU time.
Vercel's September 2025 restructuring replaced the fixed-allowance model with credit-based billing and moved several Enterprise-only features down to Pro, as detailed on the Vercel Blog. The change affected how bandwidth, compute, and build minutes are tracked. It was the third major pricing revision in two years.
Credit-based billing. Pro plans now include a $20/month flexible spending credit that offsets usage charges. Previously, you had fixed allowances (like 1TB bandwidth) and only paid when you exceeded them. The credit system applies to compute and bandwidth overages but not to add-on subscriptions.
Enterprise features moved to Pro. Several features that previously required Enterprise -- including advanced deployment protection and some observability tools -- became available on Pro as paid add-ons. This lowered the barrier for mid-size teams but added more line items to Pro bills.
Build minute pricing restructured. The old model included 3,000 build minutes on Pro. The new model charges per-minute based on machine tier, with the $20 credit absorbing small usage. Turbo machines becoming the default for new projects in February 2026 amplified this change.
We've tracked Vercel's pricing page via the Wayback Machine since 2023. Each restructuring adds billable dimensions while adjusting base allocations. The September 2025 change looked like a simplification on paper, but credit-based billing actually makes cost prediction harder because credits apply to some charges but not others.
Vercel's September 2025 pricing update introduced credit-based billing with a $20/month flexible credit on Pro plans and moved several Enterprise features to Pro as paid add-ons, as covered on the Vercel Blog. Pricing documentation was last updated February 27, 2026.
Related: Vercel Pro pricing breakdown for teams
Preview deployments don't appear as a separate line item on your Vercel bill, but they consume build minutes, bandwidth, and function invocations from your shared Pro allowance as explained in the Vercel preview deployment docs. For active teams, preview builds can become the single largest cost driver -- especially with Turbo machines enabled by default.
Here's the math that catches teams off guard. Since February 2026, new Pro projects default to Turbo build machines at $0.126/minute. A moderate team workflow looks like this:
| Variable | Value |
|---|---|
| Average build time | 5 minutes |
| PRs per day | 20 |
| Working days/month | 22 |
| Builds per PR | ~1.5 (initial + updates) |
Monthly preview build cost = 20 PRs x 1.5 builds x 5 min x 22 days x $0.126/min
= $415.80/month
Even a conservative estimate -- one build per PR -- comes to $277/month in build minutes alone. And that doesn't count the production builds, bandwidth from QA reviews, or functions triggered during testing.
Switch to Standard or Enhanced build machines for preview deployments. Standard machines at $0.014/minute cut that $415 bill to $46. You can configure this per-project in your Vercel dashboard under Build Settings.
Also consider enabling the "Cancel previous deployments" option. Without it, rapid pushes to the same PR trigger parallel builds that all bill independently.
In our experience tracking build logs across multiple team projects, preview deployments account for 30-50% of total build minute consumption on active repositories. Teams that don't configure build machine tiers for previews pay 5-9x more than they need to.
Vercel preview deployments consume build minutes at $0.126/minute on Turbo machines, the default since February 2026. A team running 20 PRs per day with 5-minute builds can spend over $378/month on preview build costs alone.
Related: How to set up preview environments for every pull request
Vercel deprecated its managed Postgres and KV offerings in December 2024, migrating users to Neon (Postgres) and Upstash (Redis) respectively. Vercel Blob storage remains available at $0.023/GB-month. If you're searching for "Vercel Postgres pricing," the answer is: it no longer exists as a Vercel product.
| Service | Status | Pricing |
|---|---|---|
| Vercel Postgres | Deprecated (Dec 2024) | Migrated to Neon |
| Vercel KV | Deprecated (Dec 2024) | Migrated to Upstash Redis |
| Vercel Blob | Active | $0.023/GB-month storage |
| Neon (via integration) | Active | Free tier: 0.5GB, Pro from $19/mo |
| Upstash Redis (via integration) | Active | Free tier: 10K commands/day |
If you were using Vercel Postgres, your database now runs on Neon with its own pricing and billing relationship. Neon's free tier gives you 0.5GB of storage and 0.25 compute units. Beyond that, you're paying Neon directly -- not through your Vercel bill. The same applies to Upstash Redis.
This matters for budgeting. Your "Vercel costs" now include separate invoices from database providers that aren't covered by your $20 Pro credit or Vercel's spend management tools.
Vercel deprecated its managed Postgres and KV storage in December 2024, migrating users to Neon and Upstash respectively. Vercel Blob remains active at $0.023/GB-month. Database costs are now billed separately through third-party providers.
Vercel offers three observability add-ons, all billed separately from your Pro plan: Observability Plus at $10/month base, Web Analytics at $0.00003/event, and Speed Insights at $10/project/month. These are per-project charges, not per-team -- a distinction that scales costs quickly for monorepos or multi-app setups.
| Add-on | Pricing | Notes |
|---|---|---|
| Observability Plus | $10/month base | Log retention and query limits |
| Web Analytics | $0.00003/event | ~$3 per 100K events |
| Speed Insights | $10/project/month | Per project, not per team |
| Image Optimization | $0.05/1K transformations | Changed Feb 2025 |
Speed Insights at $10/project/month seems reasonable for one app. But a monorepo with three frontend projects pays $30/month for Speed Insights alone. Four projects: $40/month. This per-project model also applies to Static IPs ($100/project/month) and some other add-ons.
Web Analytics pricing looks cheap at first -- $0.00003 per event. But a site with 500K monthly page views and 3 events per page view generates 1.5 million events per month. That's $45/month just for analytics on a single project.
| Add-on | Price |
|---|---|
| SAML SSO | $300/month |
| HIPAA BAA | $350/month |
| Advanced Deployment Protection | $150/month |
| Static IPs | $100/project/month |
These compliance and security add-ons aren't one-time fees. A healthcare startup needing HIPAA compliance, SSO, and static IPs for one project pays $550/month in add-ons before any usage charges.
Vercel's observability add-ons are priced per-project: Speed Insights costs $10/project/month and Web Analytics charges $0.00003/event, roughly $3 per 100,000 events. Enterprise add-ons for SAML SSO ($300/mo) and HIPAA BAA ($350/mo) add significant fixed costs.
Related: How to add web analytics without third-party scripts
Vercel's pricing page is thorough, but certain cost dynamics only become visible once you're running production workloads. These eight gotchas have caught teams we've spoken with off guard.
Vercel bills all served bandwidth at $0.15/GB, including DDoS traffic. A volumetric attack pushing 5TB of traffic through your site costs $600 in overages. Vercel's DDoS mitigation blocks some attacks at the edge, but traffic that reaches your application gets billed regardless.
Vercel's Spend Management feature is opt-in, not automatic. Without it, your bill scales without limit. Teams that forget to enable spend caps after onboarding can face unexpected charges from traffic spikes, bot crawlers, or misconfigured functions.
Functions are billed on CPU time, provisioned memory, and invocations simultaneously. A function using 1GB memory for 100ms costs for all three dimensions. Most competing platforms charge on a single axis (execution time or invocations). This makes function cost estimation harder without detailed profiling.
As covered above, Turbo machines at $0.126/min are the default for new Pro projects. Every preview deployment triggers a full build on these machines. Five-minute Turbo builds across 20 daily PRs can cost over $378/month in build minutes alone.
Ten developers means $200/month before any usage. Twenty developers: $400/month. There's no volume discount on Pro seats. Viewer seats are free, so read-only access for stakeholders doesn't cost extra -- but anyone who deploys or configures projects needs a paid seat.
Speed Insights ($10/month) and Static IPs ($100/month) are charged per-project. A team managing five projects pays five times the listed price. This catches monorepo teams especially hard.
Your $20/month flexible credit offsets compute and bandwidth overages only. Add-on charges like Observability Plus, Speed Insights, and compliance features bill on top. A Pro plan with $10 Speed Insights and $10 Observability Plus costs $40/seat/month effectively.
Next.js middleware runs at the edge on every incoming request. Each execution counts against your 10 million edge request allocation. A site with 15 million monthly requests exhausts the included allocation and incurs $10 in overage charges -- just from middleware.
Worth noting: gotchas 3 and 8 compound. If your middleware invokes a serverless function, every request triggers charges on edge requests, function invocations, CPU time, and memory simultaneously. A single page load can generate four separate billing events.
Related: Next.js hosting cost calculator with real formulas
A 5-person team on Vercel Pro pays between $100 and $883/month depending on traffic and add-ons, while self-hosting on a Hetzner CAX21 (ARM, 4 vCPU, 8GB RAM) costs EUR 7.49/month with 20TB bandwidth included. That's roughly 90% less -- but cost isn't the only variable worth comparing.
| Cost item | Vercel Pro (5 seats) | Self-hosted (Hetzner VPS) |
|---|---|---|
| Base / hosting | $100/month | ~$8/month |
| Bandwidth (2TB) | $150 overage | Included (20TB) |
| Build compute | $50-378/month | Included |
| Analytics | $45/month (500K views) | Included (self-hosted) |
| Error tracking | External (Sentry ~$26/mo) | Included (self-hosted) |
| Speed monitoring | $10/project | Included (self-hosted) |
| Session replay | External (FullStory ~$300/mo) | Included (self-hosted) |
| SSL | Included | Free (Let's Encrypt) |
| Total (moderate) | $355-683/month | ~$8/month |
Self-hosting isn't zero effort. You're responsible for:
For a solo developer, expect 2-4 hours per month of maintenance. Tools like Temps reduce that by automating deployments, SSL, and monitoring in a single binary -- but the underlying server is still yours to manage.
Vercel's DX is genuinely excellent. Git push to deploy, automatic preview URLs, instant rollbacks, and a global edge network are hard to replicate. If your team has zero interest in infrastructure and your bill stays under $200/month, Vercel Pro is solid value. The cost conversation only gets serious when per-seat charges, usage overages, and add-ons start compounding.
[PERSONAL EXPERIENCE] We've seen teams where a Hetzner CAX31 at EUR 13.49/month comfortably handles workloads that cost $400-600/month on Vercel Pro -- same uptime, same deployment speed, but with analytics, error tracking, and session replay included at no extra cost.
A 5-person team on Vercel Pro pays $100-683/month depending on traffic and add-ons. The same workload self-hosted on a Hetzner CAX21 costs approximately EUR 7.49/month with 20TB bandwidth included — though your team takes on server management, security patching, and monitoring that Vercel handles for you.
Related: Step-by-step guide to migrating from Vercel
According to Vercel's pricing page, Vercel Pro costs $20 per seat per month with a $20 flexible spending credit, 1TB bandwidth, and 10 million edge requests included. A solo developer's minimum bill is $20/month. A 5-person team starts at $100/month before usage overages or add-ons. There are no volume discounts on Pro seats.
No. Vercel deprecated its managed Postgres and KV storage in December 2024. Existing databases were migrated to Neon (Postgres) and Upstash (Redis). You can still provision these through Vercel's integration marketplace, but they bill through their own platforms -- not through your Vercel plan.
Preview deployments don't have a separate charge, but they consume build minutes, bandwidth, and function invocations from your shared Pro allowance. With Turbo machines as the default, 20 PRs/day with 5-minute builds costs roughly $378/month in build minutes alone. Switching to Standard machines drops that to about $46/month.
Not by default. Spend Management is opt-in through your dashboard. Without it, your bill scales without limit. DDoS traffic, bot crawlers, and function retry loops can all push costs up. Enable spend limits immediately after creating your Pro account.
Vercel Enterprise costs approximately $45,000 per year at the median, based on 63 tracked purchases. It includes a 99.99% SLA (versus 99.9% on Pro), custom resource allowances, priority support, and access to features like SAML SSO and HIPAA BAA without per-add-on pricing. Contracts are negotiated annually.
Related: Best ways to deploy a Next.js app: 5 platforms compared
Pricing data reflects Vercel's published rates as of February 27, 2026. Verify current pricing at vercel.com/pricing before making infrastructure decisions. For self-hosting comparisons, check Hetzner Cloud for current VPS rates.