Features
Local inference on Apple Silicon (M-series chips)
Model conversion and compression
Quantization schemes co-designed with architecture
Hardware-aware tensor operation optimization
Batch size = 1 inference engine
Real-time decoding for interactive AI
Apple device automation integration
Models library for pre-optimized models
Sparse buffers for KV cache (June 2026)
New quantization method for speed-quality frontier (June 2026)
Frontier-level edge intelligence with Reka Edge model
Multimodal understanding (vision, video, text)
Scalable video tagging, search, and clipping via API
Video processing without replacing existing VMS
Model Context Protocol (MCP) support for video
Inference API optimized for speed and reliability
Claru training data: egocentric video, robotics trajectories
Private model deployment for public sector
PhysicalRealismBench-U benchmark for video world models
OpenRouter integration for zero-code model switching
n8n community node for automated workflows
Vision Labs for custom vision model training
Partnership with Moonvalley for physical AI co-development
CS2-10k egocentric dataset release for research
Edge inference without cloud dependency
