#I have a question
1 messages · Page 1 of 1 (latest)
Short version, because somebody pays for the model every time it thinks.
OpenClaw itself is open source, but the things behind it usually are not:
- LLM API usage costs money
- web/search/tool calls can cost money
- hosting/gateway infrastructure costs money
- unlimited free access gets abused fast
You can run it very cheaply, or sometimes locally, but there’s usually a tradeoff in quality, speed, or hardware requirements.
If you want, I can suggest the cheapest practical OpenClaw setup right now. Docs for model/provider options: https://docs.openclaw.ai/concepts/models
So that means if I download Open Claw over the Command line, is it free for me or how?
Yep, downloading/installing OpenClaw is free.
What costs money is usually the model/provider you connect to it after install, for example OpenAI, Anthropic, Gemini, OpenRouter, etc.
So the split is basically:
- OpenClaw app: free
- Using paid cloud models through it: costs money
- Using local models on your own machine: no API bill, but you need enough hardware, and quality can be worse for agent stuff
Some providers give free credits/trials, but usually not unlimited.
If you want, I can point you to the cheapest beginner setup.
So I'm going to download Open Claw and not use it on web anymore?
Basically yes, if you install OpenClaw yourself, you’re running your own copy instead of using a hosted web version.
One important detail though:
- you might still use a browser UI locally
- but the backend is running on your machine/server
- and you still need to connect a model provider, unless you run local models
So it’s less “no web at all” and more “not relying on someone else’s web app anymore”.
Getting started docs are here: https://docs.openclaw.ai/start/faq
If you want, I can help you pick between:
- cheapest setup
- easiest setup
- fully local setup
Thank you, I'll find my way around that!