Comparison
OpenClaw Launch vs QClaw: Desktop Agent or Managed Agent?
QClaw is Tencent's consumer AI agent, and it is built on the same OpenClaw framework this site hosts. The interesting difference is not the model or the skills — it is where the agent lives. QClaw puts it on your desktop; OpenClaw Launch puts it on a server that never sleeps.
Quick comparison
| Area | OpenClaw Launch | QClaw (Tencent) |
|---|---|---|
| Underlying framework | OpenClaw | OpenClaw, with a consumer layer |
| Where the agent runs | Managed server | Your Mac or Windows PC |
| Available when your PC sleeps | Yes | No |
| Chat channels | Telegram, Discord, WhatsApp, WeChat, web chat | Messaging apps including WhatsApp and Telegram |
| Local file access | Inside the managed workspace | Direct access to your own disk |
| Setup | Browser, no install | Desktop app install |
| Scheduled and background tasks | Run on the server | Run while the machine is awake |
| Availability | Generally available | Beta with limited slots |
What QClaw is
QClaw is Tencent's consumer AI agent for Windows and macOS. It runs on your own computer and turns messaging apps such as WhatsApp and Telegram into a remote control for that machine: you send it a message, and it sends email, organises files, or works through a task on your desktop while you are elsewhere.
Tencent built it on the open-source OpenClaw framework and wrapped it in a consumer layer with preconfigured scenarios, so a non-technical user can get an agent running without touching a config file. Agent actions and file handling happen on your device, while model inference is routed to the model provider you select. It also ships a security module, called Gateway, that watches agent activity for malicious instructions.
Note the name collision before you download anything: an unrelated open-source project published under the QuantumClaw organisation also uses the name QClaw. They are different pieces of software.
What OpenClaw Launch is
OpenClaw is the same open-source agent runtime underneath QClaw. OpenClaw Launch is managed hosting for it: you configure the agent in a browser, deploy it, and it runs as a long-lived service on a server. There is nothing to install, and the agent keeps answering on Telegram, Discord, WhatsApp, WeChat, or web chat regardless of what your own machine is doing.
The real difference is the machine, not the model
Because both are OpenClaw underneath, comparing them on skills or model support misses the point. Both can call strong models. The decision is about where the agent lives, and that has three practical consequences.
Uptime
A desktop agent is awake when your desktop is. That is fine for tasks you kick off while working, and it is the wrong shape for a morning summary at 7am, a reminder while you are travelling, or a bot that colleagues message during your night. Scheduled work is where people usually notice the limit first.
Reach into your own files
This one favours QClaw. An agent running on your laptop can open the documents already on that laptop. A hosted agent works inside its own managed workspace, so local files have to be uploaded or synced. If your main use case is "tidy the folders on this machine", local execution is the better tool.
Who carries the operational load
On the desktop, your machine is the runtime: its sleep settings, disk, network, and OS updates all become agent-reliability questions. Managed hosting moves that to a server, along with restarts, upgrades, and channel reconnection.
Choose QClaw when
- The task is about files and applications already on your own computer.
- You want a packaged consumer app rather than a configured service.
- You prefer local execution for privacy reasons and accept the uptime tradeoff.
- Your machine is normally on when you would want the agent.
Choose OpenClaw Launch when
- The agent must answer when your laptop is closed.
- You want scheduled or recurring work to run reliably.
- Several people need to reach the same bot.
- You would rather not run an always-on process on your own hardware.
- You want to configure the underlying OpenClaw runtime directly instead of through a consumer layer.
Bottom line
QClaw is a genuine signal that OpenClaw has reached mainstream packaging: a major platform shipped a consumer product on top of it. If your agent's job is on your desktop, a desktop agent is the right shape. If its job is to be reachable — on a schedule, from a phone, by other people, at hours when your computer is asleep — then the agent needs to live somewhere that stays awake, and that is what managed hosting is for.
OpenClaw Launch vs QClaw FAQ
Is QClaw built on OpenClaw?
Yes. Tencent built QClaw on the open-source OpenClaw framework and added a consumer-facing layer to lower the setup barrier. That is why the two share so much vocabulary — QClaw is a packaged product around the same underlying agent runtime, not a competing framework written from scratch.
Does QClaw keep working when my computer is off?
No. QClaw runs on your Mac or Windows machine, so the agent is only available while that machine is awake and online. If you close the laptop, the agent stops responding. A managed instance runs on a server instead, which is the main practical reason people move a personal agent off their desktop.
Is QClaw free?
The international beta has been offered free with a limited number of early-access slots. Beta terms, availability, and any future pricing are set by Tencent and can change, so check the official QClaw site rather than relying on a third-party page for the current terms.
Is QClaw the same as the QuantumClaw open-source project?
No, and the name collision causes real confusion. Tencent's QClaw is a consumer desktop agent. There is also an unrelated open-source project published under the QuantumClaw organisation using the same name. Check the publisher before downloading anything.