Guides
Code review guides, model benchmarks and comparisons
How pull requests and reviews work on GitHub, what the shorthand means, how each model does as a reviewer, and how review apps compare on the same pull requests.
Pull requests and code review
Pull requests and code review
Code review termsLGTM, PTAL, nit, and which ones block a merge.Pull request checklistWhat to check before you ask for review.What is a pull requestWhat a pull request holds, how it is reviewed, and the three ways to merge.What is code reviewWhat a reviewer checks, a short checklist, and where AI review fits.How to review a pull requestHow to review a pull request on GitHub, and where reviewers miss real bugs.git request-pullWhat git request-pull prints, which warnings mislead, and when to open a PR instead.LGTM meaningWhat LGTM means on GitHub, Gerrit and Kubernetes, and when it is a rubber stamp.Code review comment examplesReal review comments that found bugs, and real ones that matched nothing.Code review skills and promptsHow to write a code review skill or prompt for Claude Code or Codex.
Models
Models
Best LLM for code reviewThe best model for each kind of review, measured.DeepSeek for code reviewThe model IonWarp runs, measured lane by lane.GLM for code reviewGLM 5.3 Flash and GLM 5.3, lane by lane.Muse Spark for code reviewMuse Spark 1.3 on code review and the specialist lanes.GPT Luna for code reviewGPT 5.6 Luna Pro, lane by lane.Gemini for code reviewGemini 3.8 Flash on the specialist lanes.
Compare
Compare
Greptile alternativeAccuracy, speed and price on the same PRs.CodeRabbit alternativeWhat a team of five pays, and what each one catches.Cursor Bugbot alternativeUsage pricing, seats and F1, side by side.Qodo alternativeQodo’s credit packs and F1, against every reviewer on the same PRs.CodeRabbit pricingEvery plan, and what five developers pay.Greptile pricingCredits per review, explained with real math.