$ git diff main -- deployment/byok.md
BYOK (bring your own key) means pointing reviews at your own Claude / Gemini / OpenAI API key so token cost is paid directly to the model vendor.5 of the 19 tracked tools genuinely support BYOK: Qodo, Claude Code Review, DeepSource, Macroscope, Git AutoReview. Crossover math on /byok-breakeven.
2+Claude Code Review - Per-seat Claude plan that includes Claude Code, with Enterprise adding usage at API ratessrc
3+DeepSource - Per-user seat on Team, billed yearly, plus pay-as-you-go AI Review credits metered per 10,000 processed lines of codesrc
5+Git AutoReview - Per-team subscription, not per seat and not per repo; BYOK on every plan, so model tokens bill to your own provider accountsrc
1
| model | input | output | source |
|---|---|---|---|
| Claude Sonnet 4.6 | $3.00 | $15.00 | anthropic.com/pricing |
| Gemini 2.5 Pro | $1.25 | $10.00 | ai.google.dev/pricing |
| GPT-4.1 | $2.00 | $8.00 | openai.com/api/pricing |
1-Key rotation: someone has to own the quarterly Anthropic / Google / OpenAI key roll
2-Rate limits: 429s and 529s under bursty PR loads need retry-with-backoff
3-Audit logging: per-developer usage attribution is your problem now
4-Cost attribution: token usage by repo / by team for chargeback
1/byok-breakeven - per-tool crossover
2/tool/claude-code-review - the managed-BYOK hybrid