Skip to main content
Verified April 2026. Cells cite each vendor’s own docs. For narrative context, see the comparisons overview. For a single tool’s detail page, click its header below.

Legend

SymbolMeaning
Yes, supported and publicly documented
No, not supported
Partial / conditional (see footnote)
Not publicly documented by the vendor

Matrix

FeatureMergeWatchCodeRabbitGreptileCopilotCodexClaude CodeBugBotQodo
Source & licensing
Open-source review logic123
AGPL / copyleft license3
Free program for OSS projects456
Deployment
Self-host available789
No seat minimum to self-host78
Air-gapped deployment
LLM flexibility
BYO LLM783
Anthropic (direct API)7
AWS Bedrock7
Local / Ollama
LiteLLM (100+ providers)
Review pipeline
Multi-agent (publicly documented)10
Parallel agent execution
Merge-readiness score (1–5)
Agent-authored PR detection
MCP server (outbound)
Conventions auto-discovery
Triggers & platforms
GitHub Checks re-run
IDE integration
GitLab
Bitbucket / Azure DevOps
Pricing
Free tier for private repos11116
Per-seat pricing12
Usage-based pricing1314121516

Footnotes

  1. Codex — the Codex CLI is open source; the hosted Codex product is closed.
  2. Claude Code — the anthropics/claude-code-action GitHub Action is open source; the Claude Code CLI/platform is a commercial Anthropic product.
  3. Qodo — the PR-Agent core is AGPL-3.0 and supports OpenAI GPT, Claude, DeepSeek, and more. The hosted Qodo Merge product is a closed commercial layer on top.
  4. Greptile — OSS and startup discounts are available but not a free tier. (greptile.com/pricing)
  5. Codex — “Codex for Open Source” program offers free API credits and ChatGPT Pro access by application.
  6. Qodo — Developer tier is $0/month with 30 free PRs/month promo and 75 credits. (qodo.ai/pricing)
  7. CodeRabbit — self-host is available only to Enterprise customers with 500+ seats; on the self-hosted build, customers plug in OpenAI, Azure OpenAI, AWS Bedrock, or Anthropic directly. (docs.coderabbit.ai/self-hosted/github)
  8. Greptile — self-hosting in AWS is available only on the Enterprise plan, which also unlocks BYO LLM. (greptile.com/pricing)
  9. Claude Code — reviews run on GitHub-hosted runners via the Action; the Claude Code CLI itself runs locally. No “self-hosted control plane” in the MergeWatch sense.
  10. Claude Code — built as a single-agent loop with tool use and a configurable --max-turns; not a pre-built multi-agent review pipeline.
  11. CodeRabbit / Cursor BugBot — 14-day free Pro trials rather than a free tier.
  12. Codex — bundled with ChatGPT Plus/Pro/Business/Edu/Enterprise plans; API access is pay-as-you-go at standard OpenAI rates.
  13. Greptile — Cloud is 30/seat/monthwith50reviewsincluded+30/seat/month with 50 reviews included + 1 per additional review.
  14. Copilot code review — non-licensed org members access it through “premium requests” that bill separately.
  15. Claude Code — Action runs cost GitHub Actions minutes + Anthropic API tokens; subscription plans are per-seat.
  16. Qodo Teams — $30/user/month annual with 2,500 credits/user/month.

How to use this matrix

  • If self-hosting is a requirement, only MergeWatch and Qodo offer it without seat gates; CodeRabbit and Greptile have enterprise-only paths.
  • If you need local/air-gapped inference, MergeWatch (via Ollama) and Qodo (via on-prem/air-gapped deployments) are the only options listed.
  • If you want a fully open-source review pipeline you can audit line-by-line, only MergeWatch offers this under AGPL-3.0 across the full pipeline (Qodo’s PR-Agent is AGPL but the hosted product is closed).
  • If you already pay for Copilot, ChatGPT, or Claude, the bundled review features are the cheapest marginal cost — but you lose multi-model choice and most have no self-host option.
See the comparisons overview for narrative context and caveats, or jump into a tool-by-tool page:

vs CodeRabbit

vs Greptile

vs GitHub Copilot

vs Claude Code

vs OpenAI Codex

vs Cursor BugBot

vs Qodo Merge

Other tools