OrcaRouter Adds AI Code Review With Token Billing, Merge Blocking
The launch shifts code review pricing from seat-based fees to token consumption and pairs it with enforced merge gates, a direct answer to cost growth and model lock-in in commercial review tools.
Reporting from 1 source: ASCII.jp.
FlashLabs launched OrcaCode Review, an AI code review feature in the OrcaRouter inference gateway. Billing is based on token usage rather than per-developer seats, with a 0% markup over provider list prices. High-severity findings fail GitHub checks and block merges until fixed. Review models are selectable from 200+ options, and the rules are open source.
OrcaCode Review runs as a two-stage recipe inside OrcaRouter. An inexpensive model screens every push in seconds, and passes without issues go to a higher-accuracy model for detailed review. The default recipe can be used as-is, or replaced with any of the 200+ models OrcaRouter handles, including different models for screening and final review.
High-severity findings (P0/P1) fail GitHub checks and block merges until fixed or explicitly approved, with findings shown as inline comments on the relevant lines. FlashLabs estimates about $53.9 per month for a five-person team handling 40 PRs monthly, based on GPT-5.5 list pricing and roughly 25,000 input and 4,000 output tokens per pass.
Rules and merge gates are published as MIT-licensed open source, with engine, prompts, and billing disclosed. A GitHub App installs in one click without repository secrets or workflow files, and a single workflow file supports in-house CI.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.