Free on one machine. Pay when you add more.
Every plan runs models on hardware you own, with no token limits. Plans differ in how many machines you connect and how many people share them.
- Chat, models and fit checks on Ollama or llama.cpp
- Live hardware dashboard with an hour of history
- VS Code coding agent and MCP both ways
- OpenAI-compatible API no token limits
- Reach it from anywhere through the encrypted relay
PlusMost popular
Your machines, shared with a small team.
$8per month
Up to 5 machines · up to 5 users
Request an inviteEverything in Free, plus
- Multiple connected devices up to 5 machines in one dashboard
- Up to 5 users in one organization
- Priority support
Pro
A larger team sharing the same GPUs.
$64per month
Up to 25 machines · 5 users included
+$15 per additional user. Volume discount above 10 users.
Request an inviteEverything in Plus, and
- Email actions send emails as a step in your tasks
- Roles and model access choose who can load, pull or only use
- Usage by member, model and machine with per-member limits
- SSO and audit log Google, GitHub or any OIDC provider
- 90 days of hardware history instead of one hour
More than 25 machines, or fully air-gapped?
Custom machine counts, the whole stack inside your network, SAML and SCIM, and a support agreement.
Compare every plan.
| Feature | Free | Plus | Pro |
|---|---|---|---|
| machines and users | |||
| Connected machines | 1 | Up to 5 | Up to 25 |
| Users | 1 | Up to 5 | 5 included, +$15 each |
| Token or usage limits | None | None | None |
| models and use | |||
| Chat, model manager and fit checks | ✓ | ✓ | ✓ |
| Ollama and llama.cpp, multi-GPU split | ✓ | ✓ | ✓ |
| VS Code agent, MCP, OpenAI-compatible API | ✓ | ✓ | ✓ |
| Image generation | ✓ | ✓ | ✓ |
| Email actions in tasks | – | – | ✓ |
| monitoring | |||
| Hardware history | 1 hour | 1 hour | 90 days |
| Usage by member, model and machine | – | – | ✓ |
| teams and security | |||
| Encrypted relay | ✓ | ✓ | ✓ |
| Shared machines across users | – | ✓ | ✓ |
| Roles and per-model access | – | – | ✓ |
| SSO (Google, GitHub, OIDC) and audit log | – | – | ✓ |
| support | |||
| Support | Docs and FAQ | Priority | Priority |
One heavy coding-agent month on a hosted API is a recurring bill. The same workload on your own card costs the electricity.
Pricing questions.
What counts as a connected machine?
A machine running the Qadenza agent and paired to your account: the desktop with the GPU, the Mac Studio, the box in the closet. The laptops, phones and editors you use to reach your models don't count, so use as many as you like.
Are there token or usage limits?
No, on any plan. Your models run on your hardware, so you're paying for the software, not per token. The only limit is what your machines can do.
How does Pro billing work?
$64 a month covers your organization, up to 25 machines and 5 users. Each user after that is $15 a month. Teams with more than 10 users get a volume discount: email us.
What's the difference between Plus and Pro for a team?
Plus lets up to 5 people share up to 5 machines. Pro is for larger teams: more machines and users, roles that control who can load or pull models, usage by member, SSO and an audit log, and email actions in tasks.
Can I change plans later?
Yes. Upgrade when you add a machine or a teammate. If you move down a plan, you choose which machines stay connected.
More than 25 machines, or no internet at all?
Email qadenzabeta@mlyth.org. We'll size a plan for larger fleets, and we deploy the whole stack air-gapped inside your network.
Your GPU, from anywhere.
Qadenza is in closed beta. Leave your email and we'll send an invite when a spot opens.