Render Traded $19 Seats for a $0.15/GB Meter: A 20-Person Team's Bill at 1TB, 2TB, and 5TB

On August 1, 2026, every legacy Render workspace was force-migrated to the plans Render announced April 23, 2026, under the headline "better pricing for fast-growing teams." The headline change is genuinely good for big teams: the Pro plan is now a $25-per-month flat subscription with unlimited members, replacing the old Professional plan's $19-per-user seat fee. A 20-person team just watched $355 a month in seat fees evaporate.

The second change got less attention: included bandwidth on Pro collapsed from 500GB to 25GB, with everything above it metered at $0.15/GB. Hobby fell further, from 100GB included to 5GB. Render says 75% of paying customers will pay less or the same under the new plans. That leaves a quarter paying more — and the question is whether you're in it. Here is the worked bill, with the seat saving and the bandwidth meter on separate lines so you can see which one is actually moving your total:

For a 20-person team, the new plan wins at every traffic level in this table — and it isn't close. But read the lines, not just the totals: at 1TB, $355 of the $358.75 saving is the dead seat fee, with the meter contributing $3.75. At 5TB the split is $355 seats versus $603.75 meter — the halved per-unit rate matters more the more you burn. The seat windfall and the bandwidth meter are two separate stories, and for smaller teams the second one runs in the opposite direction.

One scope note before the arithmetic: compute pricing (web services, workers, Postgres) is unchanged and billed on top of all of this, before and after. Everything below compares the workspace subscription plus the bandwidth layer only, holding compute constant.

What actually changed on August 1​

The old lineup charged per head. Professional was $19 per member per month with 500GB of included outbound bandwidth; Organization was $29 per member with a larger bundle. Overage was billed in 100GB increments at $30 a pop — an effective $0.30/GB, and because it moved in $30 chunks, even 1GB over the line cost the full $30.

The new lineup charges per workspace. Hobby stays free but drops from 100GB to 5GB of included bandwidth and caps out at 25 services. Pro is $25 a month flat with unlimited members, 25GB included, 15 custom domains (then $0.25 per domain per month), and self-serve SOC 2 and ISO reports. Scale is $499 a month flat with 1TB included and the enterprise compliance bundle (SSO, SCIM, HIPAA). Every plan's overage is now granular per-GB billing at $0.15/GB — exactly half the old effective rate, with no more $30 rounding cliff.

Render's own framing is that included bandwidth used to be subsidized by seat fees, and unbundling the two makes pricing fairer regardless of team size. That's half true, as the next two sections show: the per-unit meter got cheaper and fairer, while the inclusion got roughly twenty times smaller. Which half dominates your bill depends entirely on your headcount-to-egress ratio.

The worked bill, line by line​

Take a 20-person platform team on legacy Professional: the old seat line is 20 × $19 = $380 every month, traffic-independent. The new seat line is $25. Seat saving: $355 a month, before a single gigabyte is counted.

Now the meter. Old rules: the first 500GB are free, then $30 per 100GB or part thereof. New rules: the first 25GB are free, then $0.15 per GB or part thereof.

At 1TB (1,000GB). Old: 500GB over the inclusion, five $30 increments, $150 in bandwidth. Total: $530. New: 975 billable GB × $0.15 = $146.25. Total: $171.25. The meter itself is $3.75 cheaper under the new plan — the per-unit rate halving almost exactly offsets the inclusion cut at this level — and the seats contribute the other $355 of the $358.75 saving.

At 2TB (2,000GB). Old: 1,500GB over, fifteen increments, $450. Total: $830. New: 1,975 × $0.15 = $296.25. Total: $321.25. Now the meter is $153.75 cheaper too, because $0.15/GB beats $0.30/GB by more, in absolute dollars, the more you burn. Total saving: $508.75.

At 5TB (5,000GB). Old: 4,500GB over, forty-five increments, $1,350. Total: $1,730. New: 4,975 × $0.15 = $746.25. Total: $771.25. Saving: $958.75, of which $603.75 comes from the cheaper meter and $355 from the seats.

Two footnotes the totals hide. First, Scale doesn't rescue heavy bandwidth: at 5TB, Scale is $499 plus 4,000 overage GB at $0.15 = $1,099, which loses to Pro's $771. Scale buys compliance features, not a cheaper byte — check that before upgrading on bandwidth grounds. Second, there is exactly one traffic band where a 20-person team pays more under the new plan: between 25GB and roughly 975GB a month, the inclusion cut outweighs the rate cut on the bandwidth line itself — but the $355 seat saving buries that effect at every point in the band. A 20-person team cannot lose here. Smaller teams can, which is the next section.

Who actually pays more​

Shrink the headcount and the seat saving shrinks with it, while the meter doesn't care how many humans you have. The $355 monthly cushion that protects the 20-person team becomes $13 for a two-person team (2 × $19 = $38 old vs $25 new) — enough to cover about 86GB of metered bandwidth. Past roughly 112GB a month, the two-person team pays more than before.

Concretely, a three-person team serving 300GB a month: old bill $57 (3 × $19, inside the 500GB inclusion, zero overage). New bill $25 plus 275 × $0.15 = $41.25, for $66.25 total. That's a $9.25 increase — about 16% — on a team doing nothing differently. The plan got cheaper and their bill went up, because what they consumed was inclusion, not seats.

Hobby absorbs the harshest version of the same trade. A side project pushing 50GB a month paid $0 under the old 100GB inclusion; under the new 5GB inclusion it pays 45 × $0.15 = $6.75 a month. A Hobby app using the full old 100GB now pays $14.25. Nobody's startup dies over $7–14 a month, but it reprices the "free tier that actually served small apps" into a meter that starts almost immediately — the same pattern as Fly.io adding snapshot and inter-region meters to previously bundled usage in early 2026, and the same direction the whole metered-PaaS market has been walking all year.

The general rule: the new plans reward high-headcount, low-egress teams most and punish low-headcount, moderate-egress teams hardest. Render's 75% figure is plausible precisely because most paying workspaces sit in the first camp. Know which camp you're in before you celebrate or mourn — the answer is one division: monthly egress divided by seat count.

The same traffic on owned hardware​

Now price the three traffic levels against the alternative this blog keeps returning to: a small fleet of owned Hetzner-backed machines under Cluster API, where bandwidth is a capacity envelope, not a meter.

The shape that matters: Hetzner Cloud servers include 20TB of traffic per server per month at every size that can run a workload node, with additional traffic around €1 per TB. A minimal highly available fleet — three worker nodes plus a control plane and a load balancer — lands roughly in the tens of euros per month, flat, before and after your traffic grows. Map the three levels onto it:

• 1TB, 2TB, 5TB a month: all inside a single node's 20TB inclusion, let alone three nodes'. Marginal bandwidth cost at every level: effectively €0. The fleet costs the same number in the quiet month and the launch month.

Against Render's new plan-plus-bandwidth totals ($171 / $321 / $771 for the 20-person team, compute excluded on both sides), the owned fleet costs roughly a quarter of the 1TB figure and the gap widens with traffic — because one side's cost curve has a slope ($0.15 times every gigabyte, forever) and the other side's is flat until you cross 20TB per box.

That is not a claim that owned hardware is free. It is a claim about which kind of number your bandwidth shows up as: a variable line item that scales with success, or a sunk capacity envelope you provision once. The honest price of the envelope is operational — you run the control plane, the upgrades, the 3 a.m. pages — which is exactly the labor the PaaS fee buys back. Teams should weigh $771 a month against that labor with eyes open, not discover the meter only when the traffic spike invoice lands.

There is a middle path worth naming for teams that feel both pulls: keep the app on the PaaS, but stop serving bytes through its meter. A static-asset or image-heavy workload fronted by object storage with zero-egress pricing (Cloudflare R2 being the canonical example) can cut metered egress by the share of traffic that was never dynamic in the first place — often the majority. Measure your dynamic-vs-static split before migrating anything; it determines whether the meter is a rounding error or the whole argument.

The pattern to watch​

Step back and the August rollout is a clean instance of the 2026 PaaS pricing playbook: remove the seat tax that punished growing human teams, then recover the margin on the machine-metered resource that scales with usage. Vercel raised bandwidth overage to the same $0.15/GB neighborhood; Fly.io found two previously bundled categories to meter within sixty days. The seat fee was visible and resented; the egress meter is quiet and compounding. Render deserves credit for the genuinely fairer half — granular per-GB billing with no $30 rounding cliff is strictly better metering — but fairer metering of a 20-times-smaller inclusion is still a smaller inclusion.

So the decision rule for a team reading its post-migration invoice: divide monthly egress by seat count. Small teams feel the meter first — once a handful of seats serve hundreds of gigabytes, the seat saving stops covering it. A three-person team serving 400GB pays $81.25 under the new meter versus $57 before; a solo dev serving 150GB goes from $0 on Hobby (or $19 on Professional) to $21.75–$46.75. Recompute your bill under the new rules instead of assuming the flat fee saved you money. Big teams are exempt from the worry: twenty seats buy a $355 cushion no traffic level in this table can overcome. And if your traffic curve points up and to the right, price the flat envelope of owned hardware at least once. The meter only ever moves one direction.

Render's new plans are live for all workspaces since August 1, 2026 — figures above come from Render's new workspace plans docs and the April pricing announcement; Hetzner figures are per-server included-traffic and overage list prices at publish time. Bex.co is the open-source, AI-native Render alternative — push a git repo, get a running HTTPS service on machines you own, with bandwidth priced like hardware instead of metered like SaaS. Star the repo on GitHub or deploy your first app today.