- Build AI workflows for agents, coding, analysis, and more with open models
- Private by default
- Use hardware you already own
- Built-in memory management for your device
- Model recommendations tailored to your device
- OpenAI-compatible API
- Desktop GUI
Pricing
Start locally for free.
Scale with devices.
Multiplayer
Create your own mini AI cloud with a Multiplayer group
$10 / mo
$8/mo per additional connected device
Join the betaEverything in Free, plus:
- Public URL to access your local AI from anywhere
- 1 additional connected device included
- Unlimited additional devices for $8/mo each
- Run larger, smarter models across connected devices
- Run more models at a time
- Zero data retention
- No usage limits, no token costs
- Burst to cloud as needed
Cloud
Burst into cloud AI when you need to
Per token
Varies by model
- Save on token pricing with Aquaduck's public inference network
- Optionally offset token prices by contributing your idle compute
- Bring your own key for other cloud providers
Need help? Contact support at team@aquaduck.ai
Private AI deployment.
For teams and enterprise.
Coming soon
Schedule a demoEnterprise
Custom
Turn your organization's devices into a private intelligence network.
- Private deployment within the enterprise environment
- Access to smarter, larger models across existing organization hardware
- Single sign-on (SSO)
- Centralized billing and audit trails
- Multi-device management
- Role-based access controls
- 24/7 priority support
Questions?Answered.
Local AI is private by default. When running models across devices, Aquaduck encrypts data in transit and keeps sensitive account and billing information off connected devices. For organizations with stricter requirements, private Aquaduck deployments can run entirely on company-owned hardware.
For Multiplayer groups, the group administrator manages all billing and enrolled devices.
Aquaduck turns idle compute into a single intelligence network. Contributors earn from sharing idle compute, while builders burst into it as needed. Contributing is optional. Specialized hardware is not required, and most personal computers are eligible to contribute. It's a new approach to AI inference that aligns everyone who helps power the network.
Aquaduck monitors network health, balances devices, and manages recovery automatically. Devices that go offline are removed from routing within seconds, and workers reconnect after network interruptions. If a device drops during an in-flight request, the request fails fast so it can be retried immediately.