Introducing OrcaCode Review: A Unique Token-Based AI Code Review Solution

FlashLabs Launches OrcaCode Review

FlashLabs Inc., headquartered in Chiyoda, Tokyo, has announced the launch of OrcaCode Review, a new AI-powered code review solution integrated into their OrcaRouter platform. Unlike traditional monthly seat-based billing, OrcaCode Review employs a token-based model, which allows developers to pay only for the tokens consumed during the review process. This innovative approach grants developers the flexibility to choose from over 200 Large Language Models (LLMs) and generative AI models as they conduct code reviews.

Key Features of OrcaCode Review

1. Token-Based Billing: Unlike conventional services that charge per developer seat, OrcaCode Review charges based on the actual number of tokens consumed. This means that teams of varying sizes can optimize their costs based on real usage without any markup.

2. GitHub Integration: The tool is designed to automatically fail the GitHub checks for critical issues classified as P0/P1 until the issues are fixed or explicitly approved, thus blocking merges for unresolved issues. Inline comments are provided directly in the code, making it easy for developers to understand the required changes.

3. Choice of AI Models: Teams can select from over 200 models for their review process. They can easily implement a two-stage review workflow that starts with cheaper models for initial screening, followed by a final review using more accurate models, all without being locked into a specific vendor.

4. Cost-Efficiency: For an estimated usage scenario involving a five-member team conducting 40 pull requests (PRs) per month, the estimated monthly charge is around $53.90, based on current token rates for GPT-5.5 models. This is based on an approximate consumption of 25,000 input tokens and 4,000 output tokens per pass.

5. Easy Installation: OrcaCode Review can be integrated into GitHub with just one click, eliminating the need for repository secrets or workflow files, although a single workflow file can be added for teams that wish to run it in their own CI/CD pipeline.

6. Open Source: The rules and merge gates for the review process are open-sourced under the MIT license, allowing developers to inspect, use, and modify the rules as necessary.

Background on AI-Powered Code Review

AI-driven code review solutions have surged in popularity, aimed at improving the quality of pull requests while reducing the manpower required for reviews. However, many existing commercial services operate on a seat-based billing model that burdens teams financially as their size increases, leading to inflated costs unrelated to actual review needs. Moreover, model lock-in, where teams are restricted to certain vendor offerings, has been a point of contention.

What is OrcaCode Review?

OrcaCode Review is a component of the OrcaRouter platform, allowing teams to conduct code reviews based on token consumption rather than developer count. After selecting an AI model through OrcaRouter, teams can begin the review process without further subscription commitments. Reviews are executed in two phases, where all pushes are first screened using an economical model for quick feedback and, if no issues are found, are then routed for detailed scrutiny by a high-accuracy model.

For severe issues classified as P0/P1, a GitHub check fails until remediation or explicit approval is provided. Inline comments are displayed for indicated lines, and reruns can also be performed at any time.

Security Features

OrcaRouter emphasizes security. Various guardrails prevent sensitive information, API keys, or PII from being sent to the models for review, ensuring that only the necessary context and diff are forwarded. Furthermore, no review data is stored for future learning, and each call is logged for accountability, encompassing evaluation results, models used, latency, and costs involved.

Installation Options

There are two primary installation methods available:

- A simple one-click installation of the GitHub App, requiring no repository secrets or workflow files.
- An option for teams wanting to add a single workflow file to integrate OrcaCode Review into their existing systems.

Cost Structure

OrcaCode Review’s token-based billing structure contrasts sharply with traditional seat-based models that ramp up costs based on the number of developers, allowing for savings aligned with the actual review workload. The cost projections suggest significant savings for teams handling substantial review volumes, making it an appealing alternative in the market. Tokens are billed directly at provider prices with no additional markup, and teams can utilize existing provider keys or credits.

Future Enhancements

FlashLabs aims to continue enhancing OrcaRouter, striving for improved code review quality and cost transparency while expanding integrations with coding agents and CI environments.

About OrcaRouter

OrcaRouter is an AI inference gateway developed by the AI research organization, Continuum AI, and is exclusively provided by FlashLabs for the Japanese market. With its OpenAI-compatible API, users can access over 200 LLMs and generative AI models from a single gateway. Recent additions to the catalog include projections for prominent models like GPT-5.6 and Claude Opus 5, among others.