Runtime
A unified interface for executing models. Native HLLM execution is the supported path. Adapters that route the same interface to other inference engines (llama.cpp, MLX and vLLM) are in development.
Extension
Connects devices and accelerators into a coordinated computing environment. Extension is in development; support for any particular device is stated only after it has been qualified.
eGGUF
A model artifact system that keeps proprietary model representations inside Hawking's own container. eGGUF models run through Hawking's native HLLM runtime and do not require llama.cpp, MLX or vLLM. The container is sealed and verifiable, and protects its contents against casual extraction; it is not a guarantee of secrecy.
Supermodels
Models transformed through reconstruction, calibration and architecture research, aiming to improve capability per unit of resource relative to an identified source model. Supermodels are a research direction, not a released product.
Software and research code are published by Hawking Systems on GitHub ↗ as each component is qualified.