One model, one bet. Pick yours.
GitSniff lets you choose the reviewer, see the cost, and swap models per PR.
The differences that matter.
Three reasons teams move to GitSniff.
You see every dollar.
Each review logs input tokens, output tokens, and cost. No seat fee hiding what you actually spend.
You choose the reviewer.
Claude for nuance, Gemini for scale, Grok for speed. Greptile picks one model and ships it for everyone.
Chat and auto-fix ship together.
Ask questions on the PR, accept generated patches. Greptile stops at the review comment.
The full picture.
Size-aware routing
Small PRs go to a fast model. Large PRs go to a powerful one. Automatic and auditable.
Frontier models on demand
Claude Opus for architecture calls, GPT-4 for breadth, Gemini for long context. Pick per repo or per PR.
Security by default
Scanners run every PR. No separate vendor, no extra integration work.
Custom repo instructions
Drop standards into .opencode/instructions.md and every review adopts them.
No seat minimums
Quiet month, quiet bill. Scale contributors without scaling spend.
Transparent dashboards
Review cost, model choice, and findings breakdown per repo, per week.
Stop guessing which model reviewed your PR.
GitSniff shows you the model, the tokens, and the reasoning. Install the GitHub app to try it on your next PR.