Nunaki. Our new model. Local, sovereign, mobile, no cloud required. CPU-native, Edge-ready, minimal Footprint, starting at Raspberry Pi-class and up.
Technology
The Nunaki model is based on an innovative approach that enables AI inference on standard CPUs. No GPU, no compromises on quality.
10%
Power consumption vs. GPU-based AI
CPU
Runs on standard x86/ARM/RISC-V hardware
100%
Data stays local
Features
Nunaki combines natural language processing with radical data sovereignty; offline by default, online only with explicit permission.
Efficient Language Processing
Optimised NLP model with minimal latency, running directly on-device. No external API calls. Fast enough for real conversations.
Natural Language Modulation
Trained on natural conversation patterns. Nunaki doesn't respond like a machine. It responds like a conversation partner.
Offline by Default
No ping to external servers. No data in the cloud. Nunaki processes everything locally; online only on explicit user permission.
SD-Card RAG
Integrate your own knowledge content via SD card. Contextualised retrieval. Fully local, fully controllable.
Tool Calls by Permission
External tools and online access only after explicit approval. The user decides, not the manufacturer.
Offline is not a feature. It is the foundation.