Desert Ant Labs, a European AI lab, has launched its first 18 specialized models—including 12 stable models and six in beta—designed specifically for on-device intelligence. These models are optimized for audio, vision, and text tasks, aiming to provide millisecond-level responses without the costs or latency associated with cloud-based API calls.
Desert Ant Labs Launches Specialized On-Device AI Models and SDKs
The models are accessible via a single SDK for Swift, Kotlin, and JavaScript, supporting frameworks such as Core ML, LiteRT, and WebAssembly. This allows developers to implement local inference on various platforms, including iOS, Android, and the web. According to the company, the models are designed to be efficient enough to run on older mobile devices.
Desert Ant Labs provides these models for free for up to 100,000 monthly active devices per SDK, with unlimited inference per user. The company, which also develops the video app Detail, aims to reduce infrastructure costs and improve data privacy by ensuring data remains on the user's device.
Sources
- Desert Ant Labs: local, fast models that run on device （Hacker News Frontpage, 2026-09-09）
- Desert Ant Labs GitHub