aqua
duck
Purpose-built for AI agent workloads. Cost efficient, highly available edge compute with fault tolerance by design.
Optimize inference costs on long-horizon agent workloads. Ideal for high-volume, long-running agents.
High volume
Long-running agents shift the focus from latency to throughput. Coordinated edge devices with fault tolerance and multi-model routing improve supply and reduce costs on high-volume workloads.
Quick access
No infra to provision, no capacity planning. Just plug in and access low-cost, scalable compute for your agents.
distributed inference
Distributed inference and multi-model routing. A low-cost, flexible alternative for high-volume, latency tolerant workloads.
Auto-matching
Aquaduck continuously matches workloads to supply, optimizing for price, throughput, and utilization.
Auto-scaling
With auto-scaling architecture, Aquaduck manages supply and cost while agents run uninterrupted.
quack quack quack
Earn income on the spare compute from your laptop. Sign up for early access.
Earn from idle compute
Turn idle compute into passive income. Download Aquaduck and put spare memory on your laptop to work serving real AI workloads.
Easy to set up
Easy to setup on the devices you already own. Download the app and start earning as your device serves agents in the background.
benefits
- GGUF model support
- Anti-spoofing measures
- Low cost pricing
- Auto-scaling architecture
- Zero upkeep infrastructure
- Automatic node rebalancing
- Latency-aware node selection
- Mixture of Experts (MoE) sharding