site: price Compute alongside Postgres on the pricing page

Compute GA: the pricing page now covers both products as equals.

- Plan cards: symmetric labelled groups per card — Prisma Compute
  (included allowance), Prisma Postgres, and Platform (spend limits,
  compliance)
- Compare-plans table: new Prisma Compute group with included/overage
  row pairs at the GA meter rates; "Database usage" renamed
  "Prisma Postgres" and Monthly price hoisted above the product groups
- How-it-works and FAQ copy updated for GA (no more beta hedging);
  egress answer scoped to Postgres now that Compute bills bandwidth
- Meter rates added to pricing-data.ts and the llms surfaces

⚠ Compute *included allowance* numbers in the cards and table are
placeholders to establish the format — real figures TBD (rates are
the confirmed GA meter prices).

Author: sneub

apps/site/src/app/llms-content.ts

@@ -1,6 +1,6 @@
 import { SITE_HOME_DESCRIPTION, SITE_HOME_TITLE } from "@/lib/site-metadata";
 import { getBaseUrl } from "@/lib/url";
-import { plans, usagePricing } from "@/lib/pricing-data";
+import { computeMeters, plans, usagePricing } from "@/lib/pricing-data";
 
 type LlmsPage = {
   path: string;
@@ -46,9 +46,9 @@ const sitePages: LlmsPage[] = [
   },
   {
     path: "/pricing",
-    title: "Pricing — Prisma Postgres Plans & Features",
+    title: "Pricing — Prisma Compute & Prisma Postgres",
     description:
-      "Prisma Postgres plan details, included usage, database limits, support levels, and compliance coverage.",
+      "Usage-based pricing for the whole stack: Prisma Compute app hosting rates, Prisma Postgres plan details, included usage, database limits, support levels, and compliance coverage.",
   },
   {
     path: "/enterprise",
@@ -195,8 +195,8 @@ Key Prisma MCP Server capabilities:
 ## Pricing
 
 URL: ${toAbsoluteUrl(baseUrl, "/pricing")}
-Title: Pricing — Prisma Postgres Plans & Features
-Description: Get started for free with Prisma Postgres. Choose the right plan for your workspace based on your project requirements.
+Title: Pricing — Prisma Compute & Prisma Postgres
+Description: Usage-based pricing for the whole stack. Get started for free, choose the right Prisma Postgres plan for your workspace, and pay for Prisma Compute app hosting by use.
 
 Pricing summary:
 - Free plan: ${plans.free.price.USD}/month, 100,000 operations included, 500 MB storage, 50 databases, no credit card required
@@ -216,6 +216,10 @@ Support and compliance summary:
 - Pro: Standard support, GDPR / HIPAA
 - Business: Premium support, GDPR / HIPAA / SOC2 / ISO:27001
 
+Prisma Compute pricing (usage-based app hosting):
+${computeMeters.map((m) => `- ${m.meter}: ${m.price} — ${m.represents}`).join("\n")}
+- Idle apps scale to zero and cost nothing; deploys, preview branches, and seats are not billed
+
 ## Enterprise
 
 URL: ${toAbsoluteUrl(baseUrl, "/enterprise")}

apps/site/src/app/pricing/page.tsx

@@ -12,10 +12,15 @@ import { TestimonialsReveal } from "@/components/sections/testimonials-reveal";
 export const metadata: Metadata = {
   title: "Pricing",
   description:
-    "Usage-based pricing by operation — pay for the queries your app runs, not seats or deploys. Free tier with no time limit, hard spend limits on every paid plan.",
+    "Usage-based pricing for your whole stack — Prisma Compute app hosting and Prisma Postgres databases. Pay for the work your app does, not seats or deploys. Free tier with no time limit, hard spend limits on every paid plan.",
 };
 
 // Built from the approved V2 pricing copy, in V2's order (CTA then FAQs).
+// Compute GA launch (Shane, 2026-08-24): Compute pricing lives inside the
+// existing structures — included allowances on the plan cards, rates and
+// allowances as a group in the spec table — NOT as a separate section, hero
+// subhead, or strip (all three were tried and removed on Shane's feedback:
+// "we already have a table for these things").
 // "Every plan includes" is the right-hand column of PricingHowItWorks; the
 // calculator's three entry points double as its preset selector.
 //

apps/site/src/components/sections/pricing-faq.tsx

@@ -3,6 +3,12 @@ import { Faq, type FaqItem } from "@/components/sections/faq";
 // V2's pricing FAQs, copy verbatim. The operation definition is the wording
 // Gregory locked (db-only: API requests do not count) — keep it identical
 // wherever it appears on the page.
+//
+// The Compute answers were rewritten for the GA launch (Shane, 2026-08-24):
+// the beta-era "pricing will be published before general availability" line
+// became false the moment the meter table went on this page, and the egress
+// answer now scopes its "no" to Postgres since Compute bills outbound
+// bandwidth. Rates quoted here must match pricing-data.ts computeMeters.
 const PRICING_FAQS: FaqItem[] = [
   {
     question: "What counts as an operation?",
@@ -12,7 +18,7 @@ const PRICING_FAQS: FaqItem[] = [
   {
     question: "Is Prisma ORM really free?",
     answer:
-      "Yes. Prisma ORM is open source and will always be free. The paid plans cover Prisma Postgres (managed database hosting) and Prisma Compute (app hosting, currently in public beta).",
+      "Yes. Prisma ORM is open source and will always be free. The paid plans cover Prisma Postgres (managed database hosting) and Prisma Compute (app hosting).",
   },
   {
     question: "What happens if I exceed my plan's operations?",
@@ -21,7 +27,8 @@ const PRICING_FAQS: FaqItem[] = [
   },
   {
     question: "Are there egress fees?",
-    answer: "No. Unlimited data transfer is included with Prisma Postgres on all plans.",
+    answer:
+      "Not for your database — unlimited data transfer is included with Prisma Postgres on all plans. Apps hosted on Prisma Compute pay for outbound bandwidth at $0.025 per GB, as part of Compute's usage-based pricing.",
   },
   {
     question: "Can I change plans later?",
@@ -30,7 +37,7 @@ const PRICING_FAQS: FaqItem[] = [
   {
     question: "What is Prisma Compute?",
     answer:
-      "Prisma Compute is TypeScript app hosting built to run alongside Prisma Postgres — co-located on the same infrastructure so your app and database are always next to each other. It's currently in public beta. Pricing will be published before general availability.",
+      "Prisma Compute is TypeScript app hosting built to run alongside Prisma Postgres — co-located on the same infrastructure so your app and database are always next to each other. It's billed on usage across four meters — requests, provisioned memory, active CPU, and outbound bandwidth — and an idle app scales to zero and costs nothing.",
   },
   {
     question: "Is there custom pricing for high-volume teams?",

apps/site/src/components/sections/pricing-how-it-works.tsx

@@ -1,13 +1,20 @@
 import { Reveal } from "@/components/motion/reveal";
 import { PricingIncludes } from "./pricing-includes";
 
-// "How Prisma pricing works" — V2 copy verbatim on the left, the every-plan
-// -includes list stacked on the right. The calculator lives in its own section
-// after this one (see pricing-calculator.tsx).
+// "How Prisma pricing works" — the every-plan-includes list stacked on the
+// right. The calculator lives in its own section after this one (see
+// pricing-calculator.tsx).
 //
 // Sits *below* the plan cards as of 2026-07-30 (client request). Reads as the
 // explanation of the cards above it, and "Every plan includes" now lands as a
 // footnote to them, which is where it belongs.
+//
+// Rewritten for the Compute GA launch (Shane, 2026-08-24): the V2 copy
+// explained operations only, and this section is where the page's one-usage-
+// model story has to land — Compute's four meters and Postgres's operations
+// as two halves of the same idea. The operation definition keeps Gregory's
+// locked wording (db-only); the workflow-stays-free paragraph and the bold
+// spend-limit promise carry over from V2.
 export function PricingHowItWorks() {
   return (
     <section className="bg-white px-4 sm:px-8">
@@ -24,16 +31,18 @@ export function PricingHowItWorks() {
             </Reveal>
             <Reveal delay={0.1}>
               <p className="mt-6 text-pretty text-base leading-relaxed text-muted-foreground">
-                Prisma bills you for real work. Every query your app runs against your Prisma
-                Postgres database counts as one operation, and operations are what you pay for.
+                Prisma bills you for real work, on both halves of your stack. Your app on Prisma
+                Compute pays for the requests it serves, the memory and CPU it uses while running,
+                and the data it sends out. Your database on Prisma Postgres counts every query as
+                one operation, and operations are what you pay for.
               </p>
             </Reveal>
             <Reveal delay={0.15}>
               <p className="mt-5 text-pretty text-base leading-relaxed text-muted-foreground">
                 The ceremony around shipping stays free: deploys, preview branches, idle time, and
-                the number of people on your team never add to the bill. A request does the same
-                work whether it comes from a person, a script, or an AI agent, so that work is all
-                you&apos;re charged for.
+                the number of people on your team never add to the bill. An idle app scales to zero
+                and costs nothing. A request does the same work whether it comes from a person, a
+                script, or an AI agent, so that work is all you&apos;re charged for.
               </p>
             </Reveal>
             {/* V2 sets this line in bold — it's the promise the whole section

apps/site/src/components/sections/pricing-plans.tsx

@@ -9,6 +9,16 @@ import { cn } from "@/lib/utils";
 //    on the lower price rather than making Prisma read as expensive; Free
 //    states that it is free forever
 //
+// `compute` is each plan's included Prisma Compute allowance — Compute GA
+// (Shane, 2026-08-24): "put the details in the cards (how many are included
+// in each plan)".
+// ⚠ The allowance NUMBERS are placeholders: invented to show the format,
+// scaled so each plan's included Compute is roughly proportional to its price
+// at the GA meter rates ($1.00/1M requests, $0.006/GB-hour, $0.064/vCPU-hour,
+// $0.025/GB). Shane supplies the real figures — "we'll fix the exact numbers
+// after". Keep them in sync with the Prisma Compute group in
+// pricing-spec-table.tsx, which repeats them alongside the overage rates.
+//
 // `blurb` is kept in the data but NOT rendered (2026-07-30). Shane asked to land
 // on the plans with no preamble, and trimming the cards is what buys the last
 // 40px: at 1440x800 the blurb pushed each card to 494px and the CTA fell below
@@ -21,7 +31,14 @@ const PLANS = [
     name: "Free",
     price: "$0",
     note: "Free forever. No credit card required.",
-    features: ["100,000 operations / month", "500 MB storage", "50 databases"],
+    compute: [
+      "1,000,000 requests / month",
+      "100 GB-hours memory",
+      "10 vCPU-hours active CPU",
+      "10 GB outbound bandwidth",
+    ],
+    postgres: ["100,000 operations / month", "500 MB storage", "50 databases"],
+    platform: [],
     blurb:
       "Everything you need to build and explore with no clock running. No credit card, no expiry.",
     cta: "Start for free",
@@ -32,13 +49,19 @@ const PLANS = [
     name: "Starter",
     price: "$10",
     note: null,
-    features: [
+    compute: [
+      "5,000,000 requests / month",
+      "500 GB-hours memory",
+      "50 vCPU-hours active CPU",
+      "50 GB outbound bandwidth",
+    ],
+    postgres: [
       "1,000,000 operations / month, then $0.0080 per 1,000",
       "10 GB storage, then $2.00 per GB",
       "1,000 databases",
       "7-day daily backups",
-      "Spend limits",
     ],
+    platform: ["Spend limits"],
     blurb:
       "Ship your first production app without worrying about the bill. Backups included, spend limits on by default.",
     cta: "Get started",
@@ -49,14 +72,19 @@ const PLANS = [
     name: "Pro",
     price: "$49",
     note: null,
-    features: [
+    compute: [
+      "20,000,000 requests / month",
+      "2,000 GB-hours memory",
+      "200 vCPU-hours active CPU",
+      "250 GB outbound bandwidth",
+    ],
+    postgres: [
       "10,000,000 operations / month, then $0.0020 per 1,000",
       "50 GB storage, then $1.50 per GB",
       "1,000 databases",
       "7-day daily backups",
-      "Spend limits",
-      "GDPR, HIPAA",
     ],
+    platform: ["Spend limits", "GDPR, HIPAA"],
     blurb:
       "For production apps with real traffic. More operations, lower overage rates, and headroom to grow without watching the meter.",
     cta: "Get started",
@@ -67,21 +95,37 @@ const PLANS = [
     name: "Business",
     price: "$129",
     note: null,
-    features: [
+    compute: [
+      "50,000,000 requests / month",
+      "5,000 GB-hours memory",
+      "500 vCPU-hours active CPU",
+      "1 TB outbound bandwidth",
+    ],
+    postgres: [
       "50,000,000 operations / month, then $0.0010 per 1,000",
       "100 GB storage",
       "1,000 databases",
       "30-day backup retention",
-      "Spend limits",
-      "GDPR, HIPAA, SOC 2, ISO 27001",
     ],
+    platform: ["Spend limits", "GDPR, HIPAA, SOC 2, ISO 27001"],
     blurb: "Built for high-growth teams with compliance requirements and sustained high volume.",
     cta: "Get started",
     href: "https://console.prisma.io",
     popular: false,
   },
 ];
 
+// The two product groups render identically — same label style, same bullet
+// treatment — which is the "equal footing" Shane asked for (2026-08-24:
+// Compute is a first-class citizen). Compute leads, matching the
+// "Compute+Postgres" framing. Platform holds the plan-level items (spend
+// limits, compliance) so they don't masquerade as database features.
+const CARD_GROUPS = [
+  { key: "compute", label: "Prisma Compute" },
+  { key: "postgres", label: "Prisma Postgres" },
+  { key: "platform", label: "Platform" },
+] as const;
+
 // Renders bare — no <section>, no gutter, no max-width. These cards live inside
 // the hero panel (PricingHero takes them as children), which already provides
 // all three. Rendering it standalone would have no horizontal padding.
@@ -108,14 +152,36 @@ export function PricingPlans() {
               <span className="ml-1 text-base font-normal text-muted-foreground">/ month</span>
             </p>
             {plan.note && <p className="mt-2.5 text-sm text-muted-foreground">{plan.note}</p>}
-            <ul className="mt-6 space-y-3 border-t border-black/[0.06] pt-6">
-              {plan.features.map((feature) => (
-                <li key={feature} className="flex gap-2.5 text-sm leading-relaxed text-foreground">
-                  <CheckBold className="mt-0.5 size-4 shrink-0 text-prism-cyan-500" aria-hidden />
-                  {feature}
-                </li>
-              ))}
-            </ul>
+            {/* One loop for every group — Compute and Postgres get the exact
+                same treatment by construction (see CARD_GROUPS). Compute
+                numbers are placeholders — see the note on PLANS. */}
+            <div className="mt-6">
+              {CARD_GROUPS.map(({ key, label }) => {
+                const items = plan[key];
+                if (items.length === 0) {
+                  return null;
+                }
+                return (
+                  <div key={key} className="border-t border-black/[0.06] py-5 last:pb-0">
+                    <p className="text-xs font-semibold text-muted-foreground">{label}</p>
+                    <ul className="mt-3 space-y-3">
+                      {items.map((feature) => (
+                        <li
+                          key={feature}
+                          className="flex gap-2.5 text-sm leading-relaxed text-foreground"
+                        >
+                          <CheckBold
+                            className="mt-0.5 size-4 shrink-0 text-prism-cyan-500"
+                            aria-hidden
+                          />
+                          {feature}
+                        </li>
+                      ))}
+                    </ul>
+                  </div>
+                );
+              })}
+            </div>
             {/* plan.blurb is deliberately not rendered — see the note on PLANS */}
             <div className="mt-8 flex flex-1 items-end">
               {/* Brand CTAs, not the shadcn pair: the recommended plan gets