Free to run.
Priced for scale.
The agent itself is open source under MIT. Pay only if you want us to operate it for you, or if your team needs single sign-on and audit logs.
The full agent, MIT-licensed. Runs on your laptop, your VPS, your homelab.
- —All sandboxes (local, docker, ssh, singularity, modal)
- —All 12+ channel integrations
- —Bring your own keys & models
- —Community support
We host the agent. You keep the memory, the skills, the data.
- —Managed runtime · 99.9% SLA
- —Daily encrypted memory snapshots
- —1-click channel auth
- —Email support
Shared workspaces, SSO, audit logs, signed-skill registry for organisations.
- —SSO + SCIM
- —Audit log export
- —Private skill registry
- —Dedicated support engineer
Common questions.
Is Iris.Md really free?
Yes. The agent is MIT-licensed. Self-host it on anything you control and you'll never pay us a cent. Cloud and Team are conveniences on top.
Do I need to bring my own model API keys?
On Self-Host and Cloud, yes — you bring keys for the providers you want (OpenAI, Anthropic, Google, local Ollama, etc.). On Team, you can route through your existing AI gateway.
Can I switch from Cloud back to Self-Host later?
Always. We give you a one-command memory export. Your data is yours.
What about usage limits?
There are no hard task or memory limits. Cloud bills compute at cost + 10%. Self-Host is limited only by your hardware.