Cheapest Way to Host a Next.js App in 2026: Temps vs Vercel Cost Breakdown
The cheapest way to host a Next.js app in 2026 is self-hosting with Temps (~$6/mo or free). Vercel costs $0 (Hobby) → $20/seat/mo (Pro) → ~$45K/yr (Enterprise). Worked cost examples at 10K, 100K, and 1M MAU.
The cheapest way to host a Next.js app in 2026 is self-hosting on a VPS with a platform like Temps — around $6-20/month flat, versus $20/seat/month and up on Vercel Pro. Temps is the only self-hosted PaaS that bundles git-push deployment with analytics, session replay, error tracking, and uptime monitoring in a single Apache-2.0 binary, so you are not paying separately for the observability tools Vercel leaves out. Vercel itself costs $0 (Hobby, non-commercial only), $20/seat/month (Pro), or approximately $45,000/year (Enterprise median). A solo developer on Vercel pays $20/month; a 5-person team starts at $100/month. At 10K MAU expect $20–$100/month on Vercel; at 100K MAU, $120–$400; at 1M MAU, $400–$1,200/month before add-ons. Self-hosting the same workload on an $8–$20/month VPS cuts costs by 80–90%.
TL;DR: The cheapest hosting for a Next.js app is a self-hosted VPS running Temps — free to self-host, or ~$6/month on Temps Cloud (Hetzner cost plus 30%, no per-seat or bandwidth fees). Vercel Pro is $20/seat/month with a $20 spending credit, 1TB bandwidth, and compute billed at $0.128/CPU-hour. A solo developer at 10K MAU stays at $20–$60/month on Vercel. A 5-person team at 100K MAU pays $120–$400/month after bandwidth and compute overages. At 1M MAU, expect $400–$1,200/month depending on architecture. Self-hosting the same workload on an $8/month VPS cuts costs by 80–90%. Use the PaaS Tax Calculator to model your own numbers.
Vercel Pro is $20 per seat per month. A 5-person team pays $100/month before any usage overages. Enterprise contracts average around $45,000/year based on 63 tracked purchases by Vendr.
The Hobby plan hard-pauses your site when you hit limits — 100GB bandwidth, 1 million function invocations, 4 CPU-hours, 100 build minutes. There is no overage billing; you just go offline. It is also restricted to non-commercial use, which rules it out for any monetized product.
Pro adds unlimited commercial use, 1TB bandwidth, and a $20/month flexible spending credit that offsets compute and bandwidth overages. Every seat beyond the first costs another $20. Viewer seats (read-only access for stakeholders) are unlimited and free.
Enterprise pricing is negotiated. Teams with complex compliance requirements (HIPAA BAA, SAML SSO) often pay more than the median.
At 10,000 monthly active users, most Next.js applications generate:
1–5GB bandwidth per month (well within the 1TB Pro allowance)
500K–2M function invocations (within the compute credit)
5–20 build minutes per day from CI/CD activity
Solo developer: $20/month (one seat, well within all limits)
5-person team: $100/month (five seats; usage covered by the $20 spending credit)
The $20 credit on Pro absorbs most compute and bandwidth charges at this scale. Where costs climb is if you add Speed Insights ($10/project/month), Observability Plus ($10/month), or Web Analytics ($0.00003/event — about $3 per 100K events).
At 100,000 monthly active users, the cost equation shifts. Bandwidth and compute overages become real factors:
Bandwidth: A typical Next.js app with image-heavy pages averages 1–2MB per page. At 100K MAU with 10 page views each, that's 1–2TB/month — right at or just over the 1TB Pro allowance.
Edge requests: 100K users × 30 requests/session = 3M requests, comfortably within the 10M edge request limit.
Compute: Dynamic routes with SSR average 20–50ms CPU per request. At 3M requests/month, that's 16–41 CPU-hours — roughly $2–$5 in CPU charges after the credit.
5-person team cost at 100K MAU:
Line item
Cost
5 seats
$100
$20 spending credit
–$20
Bandwidth overage (1.5TB, 0.5TB over)
$75
Compute (extra CPU-hours)
$5
Build minutes (Turbo, ~1,100 min/month)
$116
Speed Insights (1 project)
$10
Total
~$286/month
The build minutes line assumes Turbo machines at $0.105/minute — the default for new Pro projects since February 2026. Switching to Standard machines ($0.014/min) cuts build costs from $116 to about $15/month.
At 100K MAU, the biggest cost lever is build machine selection, not traffic. Switching from Turbo ($0.105/min) to Standard ($0.014/min) machines saves more than all bandwidth and compute charges combined.
At 1 million monthly active users, Vercel's multi-dimensional billing creates significant costs across bandwidth, compute, and requests simultaneously.
Bandwidth math: 1M MAU × 10 sessions × 1.5MB = 15TB/month. After the 1TB Pro allowance, that's 14TB of overages at $0.15/GB = $2,100 in bandwidth alone.
Most teams at this scale either:
Move to Vercel Enterprise (custom pricing with negotiated bandwidth caps)
Migrate to a CDN + self-hosted origin architecture
Aggressively optimize page weights and implement edge caching
1M MAU realistic cost range:
Scenario
Monthly Cost
Lean app (0.5MB avg page, aggressive caching)
$400–$600
Typical app (1.5MB avg page, moderate caching)
$800–$1,200
Media-heavy app (3MB avg page, minimal caching)
$1,500–$2,500
Add full observability add-ons
+$200–$500
At 1M MAU, the bandwidth math usually forces a conversation about Vercel Enterprise or self-hosting. Enterprise contracts are negotiable; teams with predictable traffic patterns often get better rates than pay-as-you-go Pro pricing implies.
For self-hosting comparison: a Hetzner CAX41 (8 vCPU ARM, 16GB RAM) at EUR 19.49/month includes 20TB bandwidth — handling the same traffic for roughly $20/month. That's a 40–60x cost difference.
Large-project pricing on Vercel compounds across three dimensions:
1. Seat costs scale linearly. A 10-person engineering team pays $200/month in seats before any usage. Twenty engineers: $400/month. There are no volume discounts on Pro seats.
2. Build costs scale with PR activity. Large projects with active teams run dozens of PRs daily. At 50 PRs/day on Turbo machines (the default since February 2026):
50 PRs × 1.5 builds × 5 min × $0.105/min × 22 days = $866/month
Switching to Standard machines ($0.014/min) drops that to $115/month. This is the single most impactful optimization for large Next.js deployments on Vercel.
3. Per-project add-on multiplication. Speed Insights ($10/project/month) and Static IPs ($100/project/month) multiply across every project. A monorepo with 5 apps pays $50/month for Speed Insights and $500/month for static IPs — if you need them.
Typical cost for a 10-person team at 500K MAU:
Component
Monthly Cost
10 seats
$200
$20 credit
–$20
Bandwidth (7TB, 6TB over)
$900
Compute
$15
Build minutes (Standard machines)
$115
Speed Insights (1 project)
$10
Total
~$1,220/month
Use the PaaS Tax Calculator to model your specific numbers. Bandwidth is the largest variable for content-heavy applications at this scale.
Teams looking to reduce Vercel costs in 2026 most often consider Netlify, Railway, Render, Fly.io, Coolify, and Temps. Here is how the pricing models differ.
How Temps pricing works: Temps Cloud runs on Hetzner servers billed at cost plus a 30% margin — roughly $6/month for a typical setup with no per-seat fees and no bandwidth overages. Self-hosting is free under the Apache 2.0 license. There are no per-seat charges at any tier.
Netlify and Render are frequently recommended as cost-effective Vercel alternatives for teams trying to reduce per-seat pricing. Both have generous free tiers and competitive Pro plans — check their current pricing pages for exact limits and rates, as these change regularly.
Railway uses usage-based pricing rather than per-seat billing, which makes it cheaper than Vercel for large teams. At high traffic, compute and egress costs apply — see Railway's pricing page for current rates.
Fly.io is recommended for global deployment without Vercel bandwidth costs. It uses a consumption-based model where you pay for the machines you run. See Fly.io's pricing page for current per-region rates.
Coolify is an open-source self-hosting platform (Apache 2.0 licensed) that eliminates hosting costs entirely when run on your own infrastructure. Unlike Temps, Coolify does not include built-in analytics, session replay, or error tracking.
Vercel's observability stack (Speed Insights, Web Analytics, Observability Plus) adds $20–$30/project/month on top of seat costs. Teams at scale also pay separately for PostHog or Plausible (analytics), FullStory or LogRocket (session replay), and Sentry (error tracking). Those tools together cost $200–$500/month for a typical team.
Temps ships all of these as part of a single Rust binary: git-push deployment (Pingora proxy, built by Cloudflare), web analytics, session replay, error tracking, uptime monitoring, managed databases, and transactional email (SMTP, SES, and Scaleway providers built in). No add-ons, no per-seat fees, no bandwidth bills.
Self-hosting with Temps provides Vercel-like DX — git push to deploy, automatic SSL, preview environments — at VPS prices. The tradeoff: your team takes on server management (roughly 2–4 hours/month).
Vercel the company has raised over $300M in total funding and was last publicly valued at approximately $2.5 billion as of its 2022 Series D, with additional rounds reported since (per Crunchbase). For developers asking about worth in the context of pricing: the value proposition is developer experience. Git-push to deploy, automatic preview URLs per PR, global edge network, and zero-config Next.js support are hard to replicate.
The question is whether that DX premium justifies $100–$1,200+/month depending on your scale.
The cheapest way to host a Next.js app in 2026 is self-hosting on a low-cost VPS with a platform like Temps, which runs free under the Apache 2.0 license or approximately $6/month on Temps Cloud (Hetzner cost plus a 30% margin, no per-seat or bandwidth fees). That compares to $20/seat/month and up on Vercel Pro, plus separate subscriptions for analytics, error tracking, and session replay that Temps bundles in for free. Other low-cost options include Coolify (self-hosted, deployment-only, Apache 2.0) and Railway or Render (usage-based pricing, no per-seat fees), but neither includes built-in observability the way Temps does.
Vercel costs $0 (Hobby, non-commercial only), $20 per seat per month (Pro), or approximately $45,000 per year (Enterprise, median from 63 contracts tracked by Vendr). A 5-person team on Pro starts at $100/month before usage overages. The Pro plan includes 1TB bandwidth, 10M edge requests, and a $20/month flexible spending credit that offsets compute and bandwidth overages.
A 5-person team on Vercel Pro with 100K monthly active users pays ~$286/month: $100 for five seats (minus the $20 spending credit), plus roughly $75 in bandwidth overage (at ~1.5TB against the 1TB Pro allowance), $5 in extra compute, $116 in build minutes on the default Turbo machines, and $10 for Speed Insights on one project. Switching build machines from Turbo ($0.105/min) to Standard ($0.014/min) cuts the build-minutes line to about $15/month, bringing the realistic total closer to $185/month. Self-hosting the same workload with a platform like Temps on a Hetzner VPS runs approximately $15/month — no per-seat fees, no bandwidth overages, and analytics, error tracking, and uptime monitoring included in the same Apache-2.0 binary.
Vercel deployment costs depend on your plan and usage. The Pro plan ($20/seat/month) includes 1TB bandwidth and 10M edge requests. Build costs add $0.014–$0.105 per minute depending on machine tier. Turbo machines (the default for new projects since February 2026) cost 7.5x more than Standard ($0.014/min). A team running 20 PRs/day on Turbo machines can spend $231–$347/month on builds alone. See the Next.js deployment cost calculator for detailed formulas.
Premium Next.js deployment on Vercel for large projects typically costs $500–$2,000+/month. The main drivers are seats ($20 each, no volume discount), bandwidth overages ($0.15/GB over 1TB), and build minutes (up to $0.105/min on Turbo machines). A 10-person team at 500K MAU typically pays ~$1,220/month. Self-hosting the same workload on a VPS runs $20–$50/month. Use the PaaS Tax Calculator to model your exact numbers.
The cheapest alternatives depend on your team size and traffic. For teams with multiple engineers, Railway (no per-seat fees, usage-based pricing) and Render are frequently cited as cheaper than Vercel Pro. For zero hosting cost, Coolify (open-source, Apache 2.0) and Temps (Apache 2.0) let you self-host on your own VPS. Temps additionally bundles analytics, session replay, error tracking, and uptime monitoring that would otherwise cost $200–$500/month in separate SaaS subscriptions. See current pricing at each provider's pricing page for exact figures.
Vercel's last publicly disclosed valuation was approximately $2.5 billion as of its 2022 Series D, with additional funding rounds since (per Crunchbase). For product value: Vercel Pro ($20/seat/month) provides zero-config Next.js hosting, automatic preview deployments, and a global edge network. The cost-effectiveness depends on scale — at 10K MAU it is competitive; at 1M MAU, self-hosting alternatives offer 40–60x cost savings.
Vercel's Hobby plan is free for non-commercial projects. It includes 100GB bandwidth, 1 million function invocations, 4 CPU-hours, and 100 build minutes per month. When any limit is hit, your site pauses — there is no overage billing on the free plan. Commercial use requires the $20/seat/month Pro plan.
Temps provides git-push deployment with automatic SSL and preview environments (the core Vercel DX), plus built-in analytics, session replay, error tracking, uptime monitoring, managed databases, and transactional email — all in a single Rust binary. Temps Cloud costs approximately $6/month (Hetzner infrastructure cost plus 30% margin) with no per-seat fees and no bandwidth overages. Self-hosting is free under the Apache 2.0 license. Teams migrating from Vercel typically save $200–$1,500/month depending on their current observability stack.
Pricing reflects Vercel's published rates as of June 2026. Verify current pricing at vercel.com/pricing. Competitor pricing changes frequently — check each provider's pricing page for current rates. For self-hosting cost comparisons, use the PaaS Tax Calculator.