The latest release of stablyai/orca is v1.4.229. The repository description introduces it as a development environment that runs multiple coding agents together, and the provided material contains nothing about what changed in that version. So this post covers the repository's publicly documented features rather than the changes of a specific release. It is written in TypeScript and uses the MIT license.
According to the README, orca is an Electron-based orchestrator that runs agents such as Codex, Claude Code, OpenCode, and Pi in parallel worktrees. A single prompt can be replicated to five agents, and the results compared and merged. A worktree is a separate Git working space so that each task is split off and worked on independently. It can run coding agents on its own subscription and supports desktop, mobile, and remote runtimes. The companion mobile app is for receiving notifications and sending follow-up instructions on iOS and Android.
The README also has Chinese, Japanese, Korean, Spanish, French, and Portuguese translations. In the cloud push app, tests targeting failure paths — APNs/FCM session replacement, device registration deletion races, persistent push claim mix revisions — are separated by name. However, the '100x speed' found in the material is a marketing expression, and no benchmarks back it up. The features and test structure are confirmed, but the actual degree of speed improvement or stability gain cannot be judged from this material alone.