Models
Preview the leading open model families coming to Aquaduck. Join the beta to stay up to date with new releases.
Optimized for multi-device inference.
- DeepSeek
- Qwen
- MiniMax
- Z.ai (GLM)
- Kimi
- MiMo
- OpenAI (gpt-oss)
- Llama
- Gemma
- Mistral
- Nemotron
- Thinking Machines
- Ornith
Aquaduck is currently in beta. We're focused on supporting the model families builders use most today. New models are added regularly, and a complete catalog will be available soon.
Start locally, scale with devices. Burst to cloud when you need to.
Fit larger, smarter models across devices you allow. Aquaduck combines intelligent routing with model optimizations to deliver capabilities across devices.
Get early accessQuestions?Answered.
Aquaduck supports a growing set of leading open model families and fine-tuned models. During beta, we're focused on the models builders use most and will continue expanding support over time.
Yes, open models are free to run, with no usage limits or token prices. Start locally on Solo, then create a free Teams public URL for multi-device inference. If you need even more capacity, save on per-token cloud pricing through our Community Cloud. We'll soon support other cloud providers. See pricing.
Yes. Aquaduck provides an OpenAI-compatible API, making it easy to integrate with existing applications, agents, and workflows.
Yes. We're actively expanding model support and welcome requests from the community. If you'd like us to prioritize any, let us know in the Discord or by emailing us at team@aquaduck.ai.
