PLG Pricing: Examples & Companies

What is it

PLG Pricing is a sales motion where the product itself drives acquisition, expansion, and retention — typically anchored in a free or low-friction tier.

Where self-serve pricing describes how you buy (no human in the loop), PLG describes how you grow: land a single user on a free or cheap plan, convert them to paid, then expand the account as more seats and usage pile on. 219 in-corpus companies run a PLG motion, making it the single most common sales motion in the corpus. The whole pricing structure exists to make every step frictionless.

Cursor, Vercel, and OpenAI are textbook: a free or ~$20 entry that grows into a team account. Cursor runs free Hobby → Pro at $20/mo → Teams at $40/user with a token-usage pool on top. Vercel runs free Hobby → Pro at $20/month plus metered bandwidth, edge requests, and compute. OpenAI runs ChatGPT Free → Plus at $20/mo → Business at $20/seat, with a pure-usage API ($0.20 per 1M input tokens on GPT-5.x) for the developers who land bottoms-up.

The pattern originated in bottoms-up developer and prosumer software, where an individual adopts a tool before anyone in procurement has heard of it. That is exactly the shape of the AI software wave: individual developers, marketers, and analysts trying a model unaided, then dragging the tool into their team. For a deeper primer see our introduction to usage-based pricing.

How it works

PLG pricing optimises a three-step ladder: land the user, convert them to paid, then expand the account. Each step has a distinct pricing lever.

The economics only work if the expansion step delivers net revenue retention above 100% — existing accounts spending more over time through added seats and rising usage. That is why so many PLG companies in the corpus are also hybrid-priced: the seat gets a foot in the door, and the meter is what makes the account grow.

Unit math: For a hybrid PLG account, Total bill = (seats × seat_price) + (usage × unit_price). A 10-person Cursor Teams account is 10 × $40 = $400/mo in seats, plus token usage billed at $0.25 per 1M non-Auto tokens on top — the seats land the team, the meter grows the account.

For developer-API companies there is no seat to sell, so the “land” step is a free usage grant instead. Exa gives new accounts $10 in credits at onboarding plus an extra $7/month with a card on file; Modal starts with $30 of Starter credits (and up to $10,000 for qualifying startups) before its per-second GPU meter takes over. The mechanics differ, but the shape is identical: a free front door, then a meter that scales with the workload. To pick the right meter, see our guide to choosing the right usage metric.

Companies using this

219 companies in the current corpus run a PLG motion — the largest single sales-motion group, spanning developer tools (Replit AI, Codeium), prosumer apps (Midjourney, Canva), and inference/search APIs that land developers on free credits then expand on a meter (Groq, Together AI). The table below lists every one, with its plans, billing units, free-tier status, and last-verified date.

Patterns observed

• PLG ≈ self-serve plus an expansion meter. Nearly every PLG company in the corpus is also self-serve; the difference is the deliberate expansion path beyond the first purchase. Vercel is the clearest case — a free Hobby plan and a $20 Pro seat you can buy with a card, then bandwidth at $0.15/GB and edge requests at $2/1M that quietly become the biggest lines on the bill as usage grows.

• The $20 anchor is nearly universal for the prosumer tier. OpenAI Plus, Anthropic Claude Pro (technically $17), Perplexity AI Pro, Cursor Pro, and Vercel Pro all sit at roughly $20/month. It has become the reference price a PLG buyer expects for a single-user paid plan, which compresses the whole category around that number.

• The free tier is the acquisition budget. PLG vendors treat free inference as marketing spend — capped on volume, routed to cheaper models. Suno refreshes 50 credits daily on its Free plan (roughly ten songs) and gates commercial-use rights to paid tiers; Exa buys developer trials with $10 of onboarding credit. See how this ties into freemium pricing.

• Expansion runs on seats, usage, or both. Cursor expands via seats ($40/user) plus a token pool; Vercel via metered dimensions; Notion AI folded AI into a $20/seat Business plan and meters heavy agent features at $10 per 1,000 credits on top. Hybrid pricing is the PLG expansion engine.

• PLG and sales-led coexist. The common shape is PLG bottoms-up into an account, then a sales-led motion to convert it to an enterprise contract. OpenAI (Enterprise custom above Business), Anthropic (Enterprise above Team), and GitHub Copilot (Business at $19/user, Enterprise at $39/user) all layer a top-down tier on a bottoms-up funnel.

Counterexamples & variants

The counterexample is the sales-first vendor that skips the PLG funnel entirely. Harvey — deliberately excluded from this page’s company list because it has no plg motion — lands as a top-down enterprise deal at a reported ~$1,000–$2,000 per lawyer per month with a ~20–25-seat minimum. There is no free tier and no bottoms-up path. That works when the buyer is an institution with a procurement process rather than an individual practitioner, and when the product needs security review before a single user can touch production data.

A common variant is developer PLG via credits: instead of a free seat, the vendor grants free API credits and lets usage do the expanding. Modal ($30 Starter credits, up to $10,000 for startups) and Exa ($10 onboarding + $7/month) both run this shape — the “land” is a credit grant and the “expand” is rising metered spend rather than added seats. It is still PLG; the unit is just a workload, not a human.

A second variant worth flagging is the broken PLG funnel — a free tier with no expansion meter above it. A vendor that offers a generous free plan and a single flat paid tier has land-and-convert but no land-and-expand, so revenue per account is capped the moment a user upgrades. Notion explicitly moved to close this gap: when it retired the standalone $8–$10 AI add-on in 2025 and bundled AI into a $20/seat Business plan, it also added Custom Agents metered at $10 per 1,000 credits — bolting an expansion meter onto what had been a pure seat model.

What this means for buyers vs vendors

For buyers

PLG products are cheap and fast to start — but understand the expansion curve before you standardise on one. The free or ~$20 entry is engineered to grow into a per-seat-plus-usage team bill, so model where your account lands at scale, not where it starts. On Vercel, the $20 seat is rarely the line that hurts — bandwidth at $0.15/GB and edge requests at $2/1M are. On Cursor Teams, budget the $0.25/1M token pool on top of $40/user. Ask vendors for the overage rates and the historical bill of an account your size, and estimate your own with the pricing calculator.

For vendors

PLG pricing lives or dies on two numbers: free-to-paid conversion and net revenue retention. Invest the free tier as acquisition spend (cap volume, route to cheaper models), make the upgrade trigger unmistakable — Suno gates commercial rights, OpenAI gates model quality and rate limits — and build a real expansion meter, seats and/or usage. A PLG funnel with no expansion path is just a cheap subscription. For the mechanics of building that meter, see our guide to usage-based pricing models, and layer a sales-led enterprise tier on top once accounts grow past self-serve.

Explore this theme in the knowledge graph →