Render Killed Per-Seat Pricing: A Worked Five-Person Bill for Railway vs Render vs a $19 Hetzner Box

Seven months ago, the seat line item was the whole argument. Encore's February 2026 "Heroku Is Gone" survey told migrating teams that Railway's paid plans "include unlimited seats, which is a meaningful differentiator over Render and Heroku where each team member adds to the bill." Then, on April 23, 2026, Render deleted that differentiator: new Pro ($25/month) and Scale ($499/month) workspace plans moved to a flat subscription with unlimited team members included, and every legacy workspace was force-migrated onto the new lineup on August 1.

Here is the before-and-after math for a five-engineer team. Under Render's legacy Professional plan at $19 per member per month, five seats cost $95 a month before a single container booted. Today that same team pays $25 flat. Render says 75% of paying customers saw their costs decrease or stay the same under the new plans. The seat wars are over, both sides won, and seats now cost exactly $0 on Railway Pro too ($20 per workspace per month, unlimited members included).

So if seats are free everywhere, what actually decides your managed-PaaS bill in late 2026? I priced it out. Below is a worked, line-by-line bill for the same reference team on both platforms, the sensitivity analysis showing where the money really moves, and the self-hosted counterfactual: the same footprint on owned Hetzner hardware for about $19 a month.

The worked bill: five engineers, staging plus production​

The reference setup is deliberately boring: the median small-SaaS footprint. Production runs one web service (1 vCPU, 2 GB RAM), one background worker (0.5 vCPU, 512 MB), and managed Postgres at roughly 1 GB of RAM with 10 GB of storage. Staging mirrors it smaller: one Starter-class web service and one small Postgres. Five engineers all need dashboard access. Everything runs 24/7, and egress stays under 25 GB a month (we will break that assumption in the next section).

On Render, the bill is a fixed menu. The Pro workspace is $25 flat with unlimited members. Compute comes in tiers: the production web service maps to Standard at $25 (2 GB RAM, 1 CPU), the worker and the staging web service each map to Starter at $7 (512 MB, 0.5 CPU), production Postgres maps to Basic-1GB at about $20, and staging Postgres maps to Basic-256MB at $6. Storage beyond the included gigabytes bills at $0.30 per GB per month, which adds pennies at this scale.

On Railway, the bill is a meter. The Pro workspace is $20 per month and includes $20 of usage credit, with CPU metered at $20 per vCPU, memory at $10 per GB, and volumes at $0.15 per GB (monthly equivalents of per-second rates). Score the same footprint: the production web service draws about $40 (1 vCPU plus 2 GB), the worker about $15, production Postgres about $21.50 (roughly 0.5 vCPU, 1 GB RAM, 10 GB volume), the staging web service about $15, and the staging database about $6. Total usage lands near $97.50, which after the $20 included credit bills as $20 base plus $77.50 in metered usage: about $97.50 all-in.

The headline: $90 versus $97.50, roughly a wash. Neither platform wins on price at this footprint, and neither charges you a cent for the five humans. The $95 seat line item that used to dominate this exact comparison is simply gone. What decides the bill now is the shape of your compute, which is where the sensitivity analysis earns its keep.

Where the money actually goes now​

With seats at zero on both sides, three variables move the total. Team growth is conspicuously not one of them: adding engineers six through ten changes nothing on either invoice, which is precisely why Render framed the restructure around teams "growing and shipping faster with AI." To make the death of the seat line item concrete, score just the workspace fee — before any compute — at three team sizes, new flat plans versus Render's legacy per-seat Professional plan:

At twenty engineers, the old seat math alone cost more than our entire worked compute bill. That is the world Encore's February survey described — and as of the August 1 force-migration, it no longer exists.

Bandwidth is the first real wedge. Render's Pro workspace includes 25 GB of egress, then charges $0.15 per GB; Railway charges a flat $0.05 per GB with no included allotment. At 100 GB of monthly egress, Render adds (100 − 25) × $0.15 = $11.25 while Railway adds 100 × $0.05 = $5. Below ~25 GB the difference is noise; past a few hundred gigabytes, Railway's 3x-cheaper egress rate opens a genuine gap, and Render's included-allotment cut (down from the legacy plans' far larger bundles, previously subsidized by seat fees) is the quiet price of "no seat fees."

Idle staging is the second wedge, and it favors the meter. Railway bills per second, so a staging environment stopped nights and weekends costs roughly a third of its always-on price — about $7 instead of $21 for our staging pair, pulling Railway's total into the low $80s. Render's tiered compute can be suspended too, but the platform's mental model is fixed monthly tiers, so teams tend to leave staging burning. If your staging genuinely idles two-thirds of the week, per-second metering is worth $10–15 a month here.

Compute shape cuts the other way. Notice that the production web service costs $25 on Render's Standard tier but $40 on Railway's meter — bundled tiers beat the meter for chunky, always-on services, while the meter wins for small and fractional ones. A footprint of many small services favors Railway; a footprint of a few fat ones favors Render. Add Render's $0.25-per-domain custom-domain overage if you run multi-tenant vanity domains, and none of these lines individually exceeds ~$25 — the bill is now a pile of small compute lines with no single dominant item.

The self-hosted counterfactual: about $19​

Now price the same footprint on hardware you rent by the month instead of by the seat, the vCPU, or the gigabyte. Hetzner's CPX22 cloud server — 2 shared vCPUs, 4 GB of RAM, 80 GB of SSD, 20 TB of transfer — costs €7.99 a month, roughly $9.50 after the April 2026 price adjustment. Two of them, one for production (app, worker, and Postgres on one box is plenty at this scale) and one for staging, total about $19 a month. That is one-fifth of either managed bill, with twenty terabytes of transfer making the bandwidth wedge above irrelevant.

The honest accounting does not stop at $19. The managed $90 buys you managed Postgres with backups and point-in-time recovery, zero-downtime deploys, preview environments per pull request, and freedom from 3 a.m. pages about a full disk. Self-hosting buys those back with your own time: you are the DBA, the backup cron job, and the on-call rotation. For a five-engineer team shipping product, that time almost always costs more than the $70-a-month spread — which is exactly why the managed platforms can price compute at a 4–5x multiple of raw VMs and still be the rational choice.

But the counterfactual matters for a different reason: it sets the ceiling. Every dollar above ~$19 is the premium you pay for not operating the box. When seats were $95, that premium was mostly a people tax. Now it is purely an ops-convenience fee, and you can decide with clear eyes whether the convenience is worth 5x the hardware. Teams with an ops-inclined engineer, predictable workloads, and no need for preview environments per PR increasingly answer no — and that is the cohort self-hosted PaaS tooling is built for.

What to actually choose​

If you are one of the Heroku-or-bust teams Encore surveyed back in February, update your priors: the "unlimited seats" differentiator that pointed you at Railway now points nowhere, because Render matched it. Choose on compute shape instead. Steady, chunky, always-on services with modest egress fit Render's tiered menu; many small services, spiky staging, or heavy egress fit Railway's meter. Either way, budget ~$90–100 a month for the reference footprint and stop modeling headcount into the infrastructure line.

And keep the $19 number taped to the monitor. The moment your managed bill crosses 5–10x the raw-hardware cost of your footprint, you are no longer paying for hosting — you are paying for not thinking about hosting. Sometimes that is exactly the right trade. Just make sure it is a decision, not a default.

Sources: Render, "Better pricing for fast-growing teams," April 23, 2026; Render pricing; Railway pricing; Encore, "Heroku Is Gone, Here's Where Developers Are Going," February 8, 2026; Better Stack, "Hetzner Cloud Review 2026". Prices checked September 2026.

Bex.co is the open-source, AI-native Render alternative — push a git repo, get a running HTTPS service on machines you own. Star the repo on GitHub or deploy your first app today.