Desert Ant Core

Desert Ant Labs is an on device AI lab that builds small, specialised models, each designed to do one job well, and ships them through a single native SDK that adds them to an app in a few lines of code. Because the models run on the phone or computer the end user already owns, there is no cloud bill and no per call or token cost, and data stays on the device. The catalogue is aimed at app developers who want fast, private features without running inference servers. It includes Voz for speech recognition, which the company says transcribes ten minutes of audio in two seconds on an iPhone; Clear for speech enhancement; Align for word timestamps; Uhm for filler word detection; Clips for selecting highlight clips from video; Ear and Tongue for spoken and written language identification; Redact for PII redaction; Gist for topic tagging; Title for suggested titles and descriptions; Emo for emoji suggestions; and Shapes for turning rough sketches into clean shapes. Moderator for nudity detection, Schemer for typed JSON extraction and Toxic for hate speech triage are in beta. The SDK is on GitHub, the models are on Hugging Face, and documentation is public. Every model is free up to 100,000 monthly active devices per platform, with no limit on how often each person runs it, and no login or token purchase is needed to start. Pricing above that threshold is not published. The company describes its role as the fast, local layer that sits alongside cloud reasoning models, sending work to the cloud only when a task needs it.