NeuronForge launches real-time edge LLM for mobile apps after $30M seed

AI · 4 min read

NeuronForge revealed a $30 million seed round led by Atlas Capital and launched ForgeLite, an edge-optimized language model designed specifically for mobile applications. The model is tuned for conversational latency and privacy, running directly on modern mobile NPUs.

ForgeLite targets messaging, personal assistants, and content generation inside apps, offering developer tools to quantize and prune models to meet specific device profiles. The SDK includes privacy-preserving features like on-device prompt caching and differential privacy hooks.

Investors say the funding will scale engineering, pay down licensing for low-level hardware kernels, and expand partnerships with phone OEMs for tight hardware acceleration. NeuronForge will also open a beta for enterprise mobile customers to test on-device moderation and summarization.

Analysts call the combination of funding and a tangible edge product notable, because it reduces cloud costs and regulatory exposure for apps handling sensitive user content while still offering modern LLM functionality.