via “cross-platform on-device llm inference with hardware-agnostic abstraction”
Run frontier LLMs and VLMs with day-0 model support across GPU, NPU, and CPU, with comprehensive runtime coverage for PC (Python/C++), mobile (Android & iOS), and Linux/IoT (Arm64 & x86 Docker). Supporting OpenAI GPT-OSS, IBM Granite-4, Qwen-3-VL, Gemma-3n, Ministral-3, and more.
Unique: Plugin-based hardware abstraction layer (Layer 5) decouples model inference from hardware implementation, enabling day-0 support for new models and NPU architectures without SDK recompilation. CGo bridge (Layer 4) provides zero-copy memory management across language boundaries, critical for mobile/IoT where memory is constrained.
vs others: Supports NPU inference natively (Qualcomm, AMD, Intel) unlike Ollama or LM Studio which focus on GPU/CPU, and provides mobile SDKs (Android/iOS) that competitors lack, making it the only true cross-device inference framework.