Is OpenBot Free? Pricing, Hosted Servers, and License Explained
OpenBot's desktop app is free under PolyForm Noncommercial. Here's what it costs to run, what hosted servers add, and who needs a commercial license.
What does OpenBot actually cost?
The OpenBot app itself is free, with no locked features behind a paywall. You download it for Mac, Windows, or Linux, connect the AI providers you already pay for (ChatGPT, Claude, Gemini, Grok, Cursor, and others), and your only ongoing spend is whatever those existing plans or API keys already cost you. If you want an always-on agent that keeps working while your laptop is closed, OpenBot offers hosted servers starting at 20 to 25 euros a month, plus a managed “teams” option for companies that want the whole setup configured around their tools. The free app runs under an open-source license with a catch: commercial use requires a separate paid license, since it’s released under PolyForm Noncommercial 1.0.0.
TL;DR
- OpenBot’s desktop and mobile apps cost nothing to download or run, and there are no feature gates behind a subscription inside the app itself.
- You bring your own model access, connecting providers like Codex (ChatGPT plans), Claude Code, Gemini, Grok, Cursor, Klein, and others instead of paying OpenBot for inference.
- Hosted always-on servers start at 20 to 25 euros ($25) a month for people who want agents running continuously without keeping a personal machine powered on.
- A managed teams offer exists for companies that want OpenBot run and configured around their existing tools rather than self-hosted.
- The source code is public on GitHub, but it ships under the PolyForm Noncommercial 1.0.0 license, meaning free use is limited to personal projects, study, and research.
- Any commercial deployment needs a separate license from the project, so businesses should factor that into planning before building workflows on top of it, not after.
- The free app comes with no per-action approval system yet, since it’s a development preview, so cost planning should include the operational risk of giving an agent broad access to a real machine.
Remy doesn't build the plumbing. It inherits it.
Other agents wire up auth, databases, models, and integrations from scratch every time you ask them to build something.
Remy ships with all of it from MindStudio — so every cycle goes into the app you actually want.
How does OpenBot’s pricing model actually work?
OpenBot separates two things that managed competitors usually bundle together: the software and the model access. The app itself, available for macOS 13 or newer, Windows 10 or newer, and Linux as an app image, doesn’t charge a subscription fee. Instead, you connect accounts or API keys you likely already hold. The transcript lists Codex through a ChatGPT plan, Claude Code through a Claude plan, Gemini through a Google AI Pro or Ultra plan, Grok through a Grok account or xAI API key, and support for Cursor, Klein, Muse code, Open code Pi, Inception Mercury, and any agent speaking the agent client protocol (such as Goose or Qwen code) pointed at an OpenAI-compatible endpoint. Local models running through Ollama or LM Studio are also supported, which means a team could run a background agent at effectively zero marginal cost.
This matters because it lets you assign different providers to different agents based on budget and task. A research agent might sit on a premium plan, a coding agent on another provider, and a low-stakes background checker on a free local model. You’re not paying OpenBot a markup on any of that. You’re just paying for the plans you’d likely have anyway.
The only direct cost OpenBot charges for is infrastructure: keeping an agent’s computer running when your own machine is off. That’s the hosted server tier, priced from roughly 20 to 25 euros a month, hosted in the EU according to the source material. For companies, there’s a separate managed offer where OpenBot sets up and runs servers configured around a business’s specific tools and workflows, though exact team pricing isn’t publicized in the source material here.
Why does OpenBot need a noncommercial license at all?
OpenBot’s code is public on GitHub, which invites the assumption that it’s free to use however you like. It isn’t. The project is released under PolyForm Noncommercial 1.0.0, a license that permits reading, running, and modifying the code for non-commercial purposes, personal projects, study, and research, but explicitly restricts commercial use without a separate agreement.
This is a meaningful distinction from fully open-source licenses like MIT or Apache 2.0, which impose no such restriction. PolyForm Noncommercial falls into the “source available” category: you can inspect and run the code, but using it to generate revenue, as part of a product, or inside most business operations likely crosses a line that requires licensing it commercially from the project.
For an individual experimenting with persistent AI agents on personal projects, this changes nothing in practice. For a startup or team that wants to build internal tooling or customer-facing workflows around OpenBot, it’s a conversation that needs to happen before any serious engineering investment, not after a workflow is already load-bearing.
Is OpenBot worth it compared to a fully hosted option like Grok Bot?
- ✕a coding agent
- ✕no-code
- ✕vibe coding
- ✕a faster Cursor
The one that tells the coding agents what to build.
The honest answer depends on whether you’d rather control your own infrastructure or hand that responsibility to a managed service. Grok Bot, the SpaceX AI product this whole category is modeled after, runs each bot on its own always-on cloud computer hosted by the provider, bundled with paid Cursor or SuperGrok subscriptions. You never have to think about uptime, but you also don’t choose the model, and your data lives on infrastructure you don’t control.
OpenBot flips that trade-off. Agents run locally by default, on a desktop or home server you keep powered on, which means the sensitive parts (files, conversations, logins, browser sessions) stay on hardware you own. The cost of that control is that your machine needs to stay awake for agents to keep working, unless you pay for one of OpenBot’s hosted servers to replicate the always-on behavior.
It’s also worth being clear that “local first” doesn’t mean offline. Your chosen model provider still receives every prompt your agents send, and any web page an agent opens still goes out over the network. What stays local is the workspace, the conversation history, the files, and the browser data, not the model calls themselves.
If you already pay for multiple AI subscriptions (ChatGPT, Claude, Gemini) and don’t want another walled-garden product that ignores all of them, OpenBot’s bring-your-own-provider model is the more economical choice. If you’d rather not run or maintain any infrastructure at all and already live inside Cursor or SuperGrok, Grok Bot’s fully managed approach removes that burden entirely, at the cost of lock-in.
What should businesses know before adopting it?
Two things matter more than price for any team considering OpenBot: the license and the security model. On licensing, PolyForm Noncommercial means a company cannot simply self-host OpenBot and build a commercial product or internal revenue-generating workflow on top of it without securing a commercial license first. That’s a legal and budgeting conversation, not a technical one, and it should happen at the evaluation stage.
On security, OpenBot is explicitly described as a development preview. Agents currently run with full access to files, commands, the network, and the built-in browser after a single initial consent, without per-action approval. There’s no sandboxing equivalent to Grok Bot’s separate cloud machine isolating the blast radius from your primary device. Any team piloting OpenBot should restrict it to trusted tasks, keep real backups, and consider running it on a spare machine or dedicated server rather than a primary work laptop, especially once agents are left running unattended for long stretches via scheduled routines.
Frequently Asked Questions
Is OpenBot completely free to use?
The app is free to download and run with no locked features. Your costs come from the AI provider plans or API keys you connect (ChatGPT, Claude, Gemini, Grok, etc.) and, optionally, OpenBot’s hosted server tier if you want always-on operation without keeping your own machine powered.
What is PolyForm Noncommercial and why does it matter for OpenBot?
PolyForm Noncommercial 1.0.0 is a source-available license that allows free use for personal, study, and research purposes but restricts commercial use. Businesses that want to build products or paid workflows on OpenBot need a separate commercial license from the project.
How much do OpenBot’s hosted servers cost?
Remy is new. The platform isn't.
Remy is the latest expression of years of platform work. Not a hastily wrapped LLM.
According to the source material, hosted always-on servers start at around 20 to 25 euros a month, hosted in the EU, aimed at users who want continuous agent operation without running their own hardware. A separate managed offer exists for teams, with setup tailored to a company’s tools.
Does OpenBot replace the need for AI subscriptions like ChatGPT or Claude?
No. OpenBot doesn’t sell model access. It connects to providers you already subscribe to or hold API keys for, including Codex, Claude Code, Gemini, Grok, Cursor, and local models through Ollama or LM Studio. You still pay those providers directly.
Is OpenBot safe to run on a primary work computer?
The project is a development preview, and agents currently operate with broad file, network, and browser access without per-action approval. It’s safer to run on a spare machine or server for unattended or scheduled tasks rather than a primary device holding sensitive work.