AI code review · Pricing and re-reviews
LemonCrow vs Greptile
Greptile is a strong AI reviewer, and teams like its comments. This page compares how the two are priced and how re-reviews work.
Pricing is compared from each product’s published prices, and makes no claim about which costs less per review. Review quality is shown separately, on a public benchmark: each row names its source, its date and who scored it, and rows scored by different judges are not like for like. LemonCrow’s GitHub pull request reviewer is in early access.
Free during early access.
Bring your own model key; pay your provider directly, no markup.
Pro: $30 per seat per month, for teams, including 50 review credits per seat.
A review uses 1 (Base), 3 (Plus) or 10 (Apex) credits. Extra credits are $1 each.
Pro also includes custom review rules and unlimited seats, repositories and integrations.
A seat is an active developer: someone who authored a pull request Greptile reviewed that month.
Free during early access.
The local review CLI is free today and needs no account.
Starter is free, for individual developers: 50 credits a month, 1 active developer, unlimited repositories.
Free for qualified non-commercial open-source projects with an OSI-approved license.
50% off for pre-Series A startups with under $2M revenue in the past 12 months.
Pro has a 14-day free trial.
Reviews every push by default, re-reviewing only the changed code and carrying forward open findings.
By default Greptile reviews when a pull request opens; re-reviewing on each push is a setting.
Credits are charged per review, so with push re-review on, each push uses credits.
Any supported provider, or locally Claude Code/Codex/OpenCode.
Greptile runs the models in its cloud. Self-hosted Enterprise routes to your own provider (Anthropic, OpenAI, Azure or Bedrock).
A self-hosted Review server with identity, organization roles and audit is part of LemonCrow Enterprise, which is sales-led. The AI pull request reviewer itself is in early access.
Custom pricing. Self-host in your own infrastructure; SSO and SAML.
GitHub Enterprise Server and Enterprise Cloud with data residency; GitLab Self-Managed, Bitbucket Data Center and Gitea; Perforce with P4 Code Review.
Security and compliance reviews, a dedicated Slack support channel, custom invoicing and payment terms, and a custom DPA and terms of service.
Local review CLI today; GitHub PR reviewer in early access.
Generally available on GitHub, GitLab and Bitbucket Cloud. Auto-approve and runtime validation (TREX) are in beta.
Pricing as listed on greptile.com/pricing, checked 20 October 2026. Other Greptile facts are from its documentation and public repositories, checked October 2026. Plans change; check the source before you decide.
| Tool | Precision | Recall | F-score | Median cost / PR |
|---|---|---|---|---|
| Augment Code Review | 65% | 55% | 59% | n/a |
| Cursor Bugbot | 60% | 41% | 49% | n/a |
| LemonCrow | 47% | 51% | 49% | $0.56 |
| Greptile | 45% | 45% | 45% | n/a |
| Codex Code Review | 68% | 29% | 41% | n/a |
| CodeRabbit | 36% | 43% | 39% | n/a |
| Claude Code | 23% | 51% | 31% | n/a |
| Greptile, scored by our judge | 20% | 51% | 29% | n/a |
| GitHub Copilot | 20% | 34% | 25% | n/a |
Rows scored by LemonCrow’s judge are our own runs, with the raw report linked. Every other row is as reported by Augment Code (We benchmarked 7 AI code review tools on large open-source projects. Here are the results., 11 Dec 2025; run 14 Nov 2025, default settings, their judge). We did not re-run those tools, so the columns are not strictly like for like. “n/a”: the source published no cost.
Method. 50 pull requests from Sentry, Cal.com, Grafana, Keycloak, Discourse (Python, TypeScript, Go, Java, Ruby), the set Greptile published (AI Code Review Benchmarks 2025, Jul 2025), scored against the answer key Augment Code expanded (golden_comments @ 3f2c8ab): 97 comments count, after dropping 40 Low-severity ones. An LLM judge decides whether a tool's comment and an answer-key comment refer to the same underlying issue (not by file or line number). Low-severity answer-key comments count as neither a true positive, a false negative nor a false positive.
- Review
- sonnet · claude · depth thorough
- Judge
- sonnet · prompt
sha256 cfe685620f23 - Engine
- 0.8.1 ·
729e960 - Cost of the run
- $32.28 and 369 model calls in all; median $0.56 and 8 calls a pull request
Run on . Every comment and judgment is in the raw run report. The row marked “scored by our judge” scores Greptile’s own comments (Inline review comments by greptile-apps[bot] on the 50 benchmark fork pull requests (first review only, replies and exact repeats dropped), read with the GitHub REST API, exact text) on the same pull requests with the judge LemonCrow used, so it shows how far the two judges differ.
Greptile reports an 82% catch rate (41 of 50 pull requests) on this set (AI Code Review Benchmarks 2025, Jul 2025): it counts whether the tool commented on the one hidden bug in each of the 50 pull requests, and does not count false positives, style suggestions or unrelated comments against it. A recall-type figure on one bug per pull request with false alarms ignored; it cannot be compared with precision-aware F-scores.
One review per PR on Greptile, with its published prices. The 10 seats cost 10 × $30 = $300 a month and include 10 × 50 = 500 credits. We assume the PRs are spread evenly, 30 per developer, so each seat's 50 included credits cover that developer's reviews.
- 300 PRs × 1 = 300 credits
- Within the 500 included, so no extra credits. Seats only: $300
- 300 PRs × 3 = 900 credits
- 900 − 500 included = 400 extra × $1 = $400
- $400 + $300 seats
- 300 PRs × 10 = 3,000 credits
- 3,000 − 500 included = 2,500 extra × $1 = $2,500
- $2,500 + $300 seats
LemonCrow, same team
Your model provider’s token cost. The benchmark above reports what LemonCrow’s review cost on 50 pull requests. LemonCrow is free during early access, and this page still does not claim it comes out lower than Greptile’s credits: Greptile’s cost per review was not measured.
No comparison is one-sided. If you need any of these now, Greptile is the safer pick:
- MaturityA mature PR bot that is generally available on GitHub, GitLab and Bitbucket Cloud. LemonCrow's reviewer is in early access: a GitHub Action, and a GitLab merge request bot.
- Codebase contextFull-repository graph indexing.
- Issue tracker contextJira, Confluence, Linear, Notion and Google Docs context.
- Runtime validationSandboxed runtime validation (TREX, in beta).
- AnalyticsAn analytics dashboard.
- DeploymentEnterprise deployment options (self-hosting, SSO, GitHub Enterprise Server, GitLab Self-Managed, Bitbucket Data Center).
Greptile is mature and generally available today. LemonCrow is earlier: a local review CLI now and a GitHub reviewer in early access, built around a bill that goes to your own model provider and re-reviews that cover only what changed. If the cost of leaving AI review on for every pull request is what stops you, that is the problem we are building for.