Stellon Labs

Building tiny frontier AI models that run on edge devices

Y CombinatorSummer 2025ActiveB2BEngineering, Product and DesignArtificial IntelligenceEdge Computing Semiconductors

What they do

Problem

AI models are often too large and resource-intensive to run efficiently on edge devices like smartphones, wearables, and embedded systems.

Solution

Stellon Labs builds tiny frontier AI models, such as its open-source KittenTTS text-to-speech model, optimized to run on-device on edge hardware.

Who it's for

Developers and companies building applications for smartphones, wearables, and embedded/IoT devices that need on-device AI.

Rohan Joshi

Founder

Building Stellon Labs where we train tiny frontier models that can run on edge devices. Ex-Meta Research, CMU SCS alum

Divam Gupta

Founder

Packing intelligence in super-tiny AI models.