--- title: "Greptile drops the seat fee and charges $1 a review" description: "Customers found the meter the week of November 6, and one engineering team has already made the agent its merge gate. Whose money moves: the seat-priced holdouts, and anyone marked on Greptile's $750 million valuation." publisher: "The Inference" section: "Platforms" published: 2026-11-13T15:08:47.479Z modified: 2026-11-13T15:08:47.479Z canonical: https://theinference.org/article/greptile-drops-the-seat-fee-and-charges-1-a-review language: en keywords: "Greptile, AI Code Review, Developer Tools, Pricing, AI Agents" --- # Greptile drops the seat fee and charges $1 a review > Customers found the meter the week of November 6, and one engineering team has already made the agent its merge gate. Whose money moves: the seat-priced holdouts, and anyone marked on Greptile's $750 million valuation. ## The merge gate got a meter [Greptile](https://theinference.org/markets/companies/greptile) sells the AI agent that reviews your pull requests before they merge. This month its customers discovered the agent bills by the review: $1 per credit shared across the whole organization, one credit for a standard review, ten for an Apex review of a large change [1][2]. The [company's own pricing page states it plainly](https://shattered.io/coderabbit-pricing-vs-greptile-qodo-bugbot-copilot-2026/), and the same page drops the old seat fee: "Pro has no seat fees." [1][2] Customers found out in public the week of November 6 [3][4]. Thomas (@bentlegen) [posted an investor update](https://x.com/bentlegen/status/2118787831552410104) and asked followers to guess when the $1 charge began; Collin Qian called the bill "tough." [3][4] Founder and CEO Daksh Gupta answered the only way a usage biller can: the goal is for Greptile to block merges on bad pull requests, the correctness bar is high, and more tokens cost more, so price tracks the model. [5] ## Verification became the bottleneck The workload behind that meter is compounding. Greptile's own State of AI Coding report puts lines of code per developer at 3.5x between June 2025 and June 2026, with median pull request size up 79%, and the company says it reviews more than 1% of all pull requests opened on GitHub daily for customers including [Nvidia](https://theinference.org/markets/companies/nvidia), [Coinbase](https://theinference.org/markets/companies/coinbase), [Decagon](https://theinference.org/markets/companies/decagon) and Vanta. [1][7] [Brex's fall 2026 data](https://www.boston25news.com/news/25-fastest-growing/TYC4BLU3QU6VPINZGUUTXJYUZU/) shows why: firms that bought a coding agent were about six times as likely to add a companion service such as a sandbox, testing or hosting within six months (13.7% versus 2.3%), and about sixteen times as likely to add a sandbox or testing service specifically (3.3% versus 0.2%). [8] *Chart: **Coding-agent buyers are six times likelier to add a testing or sandbox service within six months** Column chart of Brex fall 2026 data: among firms that bought a coding agent, 13.7% added a sandbox, testing or hosting service within six months versus 2.3% of other firms, and 3.3% added a sandbox or testing service specifically versus 0.2%.* *Share of firms adding a companion service within six months, split by whether they bought a coding agent, from Brex's fall 2026 software vendor data as reported by Boston 25 News. Percentages of firms in each group, not vendor quotes. [8]* ## The last seat in town Every dedicated vendor has followed the meter except one. [1] Cursor Bugbot left its $40 seat for usage billing averaging $1.00 to $1.50 a run; Qodo sells pooled credits at about $1.67 a review; [GitHub](https://theinference.org/markets/companies/microsoft) Copilot code review draws on AI credits and, since June 1, 2026, on Actions minutes. [1] CodeRabbit alone still prices per developer: $24, $48 and $72 a month. [1] The comparison's own math gives the decision rule: a CodeRabbit Essentials seat at $24 beats Greptile's $1 review only past roughly 24 reviewed pull requests per developer per month. [1] *Chart: **The same review volume bills four ways, $400 to $667 a month** Bar chart comparing monthly code review bills for a 20-developer team opening 400 pull requests per month: Greptile standard reviews $400, CodeRabbit Essentials $480, Cursor Bugbot $500 midpoint of a $400 to $600 range, Qodo credits $667.* *Monthly bill at 20 developers opening 400 pull requests, one review each, from published unit prices: Greptile $1 standard credit, CodeRabbit $24 per seat annual, Cursor Bugbot $1.00 to $1.50 average run (range shown), Qodo about $1.67 per review implied by its credit packs. Calculations from shattered.io's October 2026 comparison, not vendor quotes. [1]* On Gupta's own logic, Greptile's revenue tracks inference cost and pull-request volume, a second charge for code a model already charged to write. The reported $50 million Series B at a $750 million valuation, led by Menlo Ventures on October 19, was set under three weeks before the bill went public. [1] ## The team that dropped the second reviewer On November 6, engineer Devon Estes [described his team's month-long experiment](https://news.lavx.hu/article/devon-estes-questions-software-rules-as-ai-cuts-coordination-costs): a pull request merges when continuous integration passes, Greptile returns a clean review and the ticket owner approves, with no second engineer required on every change [6]. One month of practice, his own caveat. [6] ## Volume has to outrun the price of models Watch pull-request volume against model prices: per-review revenue only beats deflation if volume outruns it. The next reading is Greptile's next investor chart, the one Thomas promised this time. [3] The bill for AI-written code is paid twice, once to write it and once to check it. ## Takeaway Greptile now bills $1 per review credit and dropped its seat fee, tying revenue to the same inference cost that falls as models get cheaper. Every rival except CodeRabbit has already moved to usage, and the $750 million valuation was set three weeks before customers saw the meter. ## Sources 1. [Shattered.io, CodeRabbit Pricing vs 4 Rivals: $24 to $72 Seats [2026], updated Oct 25, 2026](https://shattered.io/coderabbit-pricing-vs-greptile-qodo-bugbot-copilot-2026/) 2. [Greptile, pricing page cited via Shattered.io comparison of vendor's own pricing, Oct 2026](https://shattered.io/coderabbit-pricing-vs-greptile-qodo-bugbot-copilot-2026/) 3. [@bentlegen on X, investor update post, Nov 2026](https://x.com/bentlegen/status/2118787831552410104) 4. [@collinhqian on X, Nov 2026](https://x.com/collinhqian/status/2118790994384805913) 5. [@dakshgup on X, pricing explanation, Nov 2026](https://x.com/dakshgup/status/2118859025391771976) 6. [LavX News, Devon Estes questions software rules as AI cuts coordination costs, Nov 12, 2026](https://news.lavx.hu/article/devon-estes-questions-software-rules-as-ai-cuts-coordination-costs) 7. [OpenAI, Greptile takes bug detection beyond code review with GPT-6 Astra, Oct 26, 2026](https://openai.com/index/greptile-code-review-codex-astra) 8. [Brex via Boston 25 News, The 25 fastest-growing software vendors in fall 2026, Nov 2026](https://www.boston25news.com/news/25-fastest-growing/TYC4BLU3QU6VPINZGUUTXJYUZU/) --- Greptile drops the seat fee and charges $1 a review — The Inference. Canonical: https://theinference.org/article/greptile-drops-the-seat-fee-and-charges-1-a-review