Mid-Market Access
Unlimited downloads for models up to 14B parameters. Perfect for indie developers and startups prototyping edge deployment.
Enterprise Access
Unlimited downloads for the entire marketplace, including 70B+ enterprise reasoning models. Includes prioritized hardware heuristics and premium SLAs.
Ultra-Tiny Background Agents (< 0.5B)
Load in milliseconds. Perfect for predictive typing, notification filtering, and background data sorting. Less than 100MB RAM.
Standard Application Assistants (1B - 3B)
The 'Daily Driver' for consumer apps. 500MB to 900MB RAM footprint. Perfect for customer service bots, localized AI, and offline drafting.
Heavy-Duty Reasoning Engines (8B - 14B)
Desktop-class AI for flagship smartphones. 2GB - 4GB RAM footprint. Perfect for on-device code generation, deep document analysis, and complex agentic workflows.
Vision-Language Models (VLMs)
Real-time visual processing for mobile AR, translation, and accessibility. Text and vision cores compressed to 2-bit.