HushAI Raises $45M Series B to Embed Private LLMs in Consumer Apps

AI · 5 min read

HushAI, a startup building on-device large language models optimized for privacy and low-power hardware, today announced a $45 million Series B led by Union Capital with participation from existing backers and several strategic mobile OEMs.

The company says the new capital will accelerate development of its TinyGPT derivative and a new SDK that lets mobile and embedded apps ship contextual, offline conversational features without sending user data to the cloud. HushAI positions its stack as an alternative for apps that need generative features but cannot compromise on data residency or latency.

HushAI also revealed partnerships with two major smartphone vendors to preinstall its runtime on mid-range devices and a pilot with a smart home vendor. The product roadmap includes compiler-level optimizations to bring multimodal capabilities to sub-2W processors and a marketplace for privacy-preserving app models.

For design and product teams, HushAI’s focus on UX for offline generative features will be one to watch — especially how they handle model updates, consent flows, and fallback experiences when the device is offline or constrained.