Blueiris V6 -

: By executing directly inside the core application space, the software eliminates network socket loopback delays. This ensures almost instantaneous verification of people, vehicles, and pets.

Blue Iris v6 allows users to load custom object detection models (ONNX, YOLOv5/8/10). These can be added globally in the AI settings and then assigned to individual cameras. Currently, users can select additional models like ipcam-general , ipcam-dark (for low-light scenes), ipcam-animal , and Delivery models. This level of customization drastically reduces false alerts triggered by harmless triggers like wind-blown leaves or rain.

Blue Iris v6 leans heavily into AI. While version 5 introduced CodeProject.AI integration, BI6 refines it. It seamlessly connects with local AI engines (running on your own hardware) to distinguish between: blueiris v6

: A core architectural change replaces the aging database system to improve reliability and speed when managing large volumes of footage. Integrated AI : Version 6 offers expanded native support for

Blue Iris V5 introduced "substreams," which reduced CPU usage from 80% to 15% by decoding low-res streams for motion detection while recording high-res direct-to-disk. : By executing directly inside the core application

V6 utilizes , which means hardware acceleration is more important than ever.

As of early 2026, Blue Iris v6 (including versions 6.0.x and beyond) brings several key enhancements: 1. Enhanced AI and Custom Model Support These can be added globally in the AI

The transition to version 6 focuses on refreshing the underlying architecture while maintaining the software's signature depth of customization:

: One of the most requested features is now reality. Version 6.0.1 introduced native AI processing , removing the mandatory requirement for external tools like CodeProject.AI or DeepStack. This built-in engine is optimized for speed on modern hardware and handles person, vehicle, and object detection with significantly lower latency.

This is a much more efficient workflow that saves CPU resources compared to constant scanning. V6 supports custom models (ONNX, Yolo5/8/10), allowing you to train the system to recognize specific objects like delivery vehicles, specific dogs, or "ipcam-dark" for night vision improvements.

: A major highlight is the move to a more efficient two-file database system. This rewrite aims to reduce file corruption and improve the speed of searching through large volumes of recorded footage.