AI Code Review Tools
Tools that review pull requests automatically with LLMs, usually as a GitHub/GitLab app that comments inline and suggests fixes. Status and plans re-checked on vendor pages on 2026-10-02 (unchanged from 2026-09-30 for the five original tools); vendor performance claims are not independently verified.
| Tool | Status / ownership | Pricing (vendor page) |
|---|---|---|
| CodeRabbit | Independent; vendor claims 17K customers and 6M repositories (not re-checked); GitHub and GitLab, plus IDE/CLI reviews and Slack/Discord agents | Essentials, Team and Advanced tiers per developer (annual or monthly billing); Enterprise custom; free for public repos; 14-day trial; usage-based reviews per reviewed file |
| GitHub Copilot code review | Part of github-copilot; Pro, Pro+, Max, Business, Enterprise; also VS Code, JetBrains, Xcode, Azure DevOps (preview) | Billed in AI Credits (Lite and Balanced review tiers) plus GitHub Actions minutes |
| greptile | Independent; full-codebase graph context | Free Starter (1 dev); Pro per seat with 50 credits, extra credits purchasable; free for qualified open source |
| graphite | Cursor announced a definitive agreement to acquire it on 2025-12-19; product stays independent (no closing announced as of 2026-10-02; Wikipedia still lists the deal as pending); stacked PRs plus AI reviews | Hobby free; Starter; Team per user (unlimited AI reviews) |
| Qodo (qodo) | Independent; agentic PR review with rules system, GitHub/GitLab/Bitbucket/Azure | Pro Team (up to 30 users, credit-based); Enterprise for 30+ users; 14-day trial |
| Cursor Bugbot (cursor) | Part of Cursor; reviews PRs on GitHub (incl. Enterprise Server), GitLab, Bitbucket and Azure DevOps; also via Cursor CLI/agents | Usage-based (rates on Cursor pricing page, not captured) |
| Claude Code Review (Claude Code) | Anthropic; research preview for Team and Enterprise; multi-agent review on GitHub PRs, tuned with CLAUDE.md / REVIEW.md; local /code-review command on any plan | Token-billed through usage credits (vendor docs) |
Observations
- Billing is moving to usage/credit models (Copilot, Greptile, Qodo, Bugbot, Claude Code Review) rather than flat per-seat.
- Review is converging with coding agents: Cursor cloud agents now appear inside Graphite, and Copilot review can open fix PRs (preview).
- Reviewers complement, not replace, human review; rules files (AGENTS.md-style instructions, see agents-md) steer what they flag.
Related
- Greptile - dedicated note for the Greptile reviewer
Sources
- https://www.coderabbit.ai and https://www.coderabbit.ai/pricing (accessed 2026-09-30)
- https://docs.github.com/en/copilot/concepts/agents/code-review (accessed 2026-09-30)
- https://www.greptile.com/pricing (accessed 2026-09-30)
- https://cursor.com/blog/graphite, https://graphite.com/pricing (accessed 2026-09-30)
- https://www.qodo.ai/pricing/ (accessed 2026-09-30)
- https://cursor.com/docs/bugbot (accessed 2026-10-02)
- https://code.claude.com/docs/en/code-review (accessed 2026-10-02)
- https://en.wikipedia.org/wiki/Cursor_(company) (Graphite deal pending; accessed 2026-10-02)
- Prices re-fetched 2026-10-02: coderabbit.ai/pricing, graphite.com/pricing, greptile.com/pricing, qodo.ai/pricing, docs.github.com Copilot code review page