Choose Orkas if
- You want open source and a lower price to start.
- You want ready-made specialists for slides, video and e-commerce, or Claude Code and Codex inside the team.
- You work on shared projects with other people.
Compare
The One vs Orkas
Both put a lead and a team of AI agents on your computer, with a board for the work. Choose Orkas if you want open source, a cheaper start, 16 ready-made specialists, Claude Code or Codex as teammates, and shared projects with other people. Choose The One if you want deliveries that arrive with evidence from the app's own records, a record of everything the agents did, and a Linux server for headless runs.
| The One | Orkas | |
|---|---|---|
| What it is | A main assistant hands bigger work to a working group, and each matter gets its own board | A “Commander” lead turns goals into to-dos and delegates them to specialists; each project has a to-do board with Auto-advance |
| Source | Closed source; the docs are public | Open source (MIT) |
| Platforms | macOS 15+ on Apple Silicon, Windows, and a headless Linux server | macOS 12+ (Apple Silicon and Intel), Windows 10+; Linux only from source |
| Models | The official service, ChatGPT sign-in, OpenAI, Anthropic and Gemini keys, any compatible endpoint including local models | Orkas credits, or your own keys and sign-ins with no markup (OpenAI, Anthropic, Gemini, DeepSeek, GLM, Kimi, Qwen and more), or any compatible endpoint |
| Price | No charge with your own keys or ChatGPT account; membership $20 / $60 / $200 a month with model usage included | Free with 500 credits; Lite $5.99, Pro $16.99, Max $29.99 a month |
| The team | Official templates: software, research, general, frontend, image, video, film and 3D studios, store teams, site launch; roles you edit | 16 built-in specialists (research, writing, slides, video, Office, coding, five e-commerce) plus Claude Code, Codex, OpenCode, OpenClaw and Hermes as teammates; a marketplace |
| Deliveries | Each delivery comes with an evidence pack (tests, diffs, commands, outside actions, spend) built from the app's own records; you accept, rework or give up | You review the plan first; tasks show status and ask for decisions and permissions |
| Record and undo | An append-only action log of every tool call; undo a call, a run, a matter or everything since a time | Its own docs: rollback “Not implemented. Reversing a batch is manual”; no client-facing audit trail |
| Automation | Triggers on a schedule or cron, when a folder, a web page or a mailbox has something new, or on a webhook | Hourly, daily, weekly or monthly; catches up the latest missed run within 12 hours |
| From your phone | Agent Room channel, text, while your computer runs The One | “Remote control is planned” |
| Working with others | Share a Space on the-zeroth.com | Shared projects up to 50 members, sync, published web apps (paid plans) |
Choose The One if
- You need a record of exactly what agents did, kept by the app rather than reported by the agents.
- You want each delivery to arrive with tests, diffs and spend you can check.
- You run agents on a Linux server without a screen.
Questions
- Are boards the difference?
- No. Both have them; the difference is how deliveries come back, the record the app keeps, and where the agents can run.
- Can The One use Claude Code or Codex?
- It can use Codex models through your ChatGPT sign-in and Claude models through an Anthropic key, and Claude Code or Codex can call The One's working groups through its local MCP server; it does not run them as teammates.
Facts checked on 2026-11-13 from orkas.ai, its docs and GitHub, and from The One's docs. Something changed? Tell us at support@the-zeroth.com