Vaidio 10.0.Vision AI taken further.

Vaidio 10.0 at a glance

With version 10.0, Vaidio is gearing its platform more towards large, distributed and computationally intensive vision AI environments. The focus is not only on new analysis functions. What is particularly important is how efficiently video streams are processed, how systems are scaled and how detected events are verified in everyday operations.

The release combines four central developments: higher GPU utilization, support for current NVIDIA architectures, recorded video analytics for stored material and Vaidio Vista for VLM-supported verification. There are also improvements for multi-node environments, global search, live views and camera status.

The scaling doesn’t end at around 1000 cameras. Vaidio describes a platform range from single cameras to installations with thousands of channels to distributed enterprise environments with tens of thousands of cameras. The new Unified Scalable Architecture is designed to operate such large and distributed systems more efficiently.

More video channels per GPU

Vaidio claims up to three times more efficient video processing for version 10.0. This is to be achieved through optimizations in decoding, processing and analysing video streams. Higher channel density per GPU can reduce hardware requirements and improve cost per channel analysed.

This development is particularly relevant when there are many cameras or multiple locations. Scaling does not automatically mean providing proportionally more computing power for each expansion. However, the actual efficiency gain in a project depends on the resolution, frame rate, analysis models, scene complexity and selected hardware.

New infrastructure with NVIDIA Blackwell and Grace Hopper

Vaidio 10.0 supports NVIDIA Blackwell and NVIDIA Grace Hopper. This prepares the platform for a new generation of accelerated computing infrastructure. At the same time, Vaidio is further developing the previous single container architecture into a more modular multi-container structure.

This Unified Scalable Architecture is intended to make larger installations more flexible and improve operational resilience. It’s not just maximum computing power that’s important for companies. Equally relevant are clear roles for individual services, plannable expansions and stable operation across multiple nodes or locations.

Recorded video analytics for stored material

So far, AI video analytics has often focused on the running camera stream. Recorded video analytics extends this approach to recorded and offline video footage from supported video management systems.

This allows you to search saved recordings for objects, people, vehicles, behaviour patterns or events. This can significantly reduce manual screening during investigations, compliance checks, follow-up analyses and operational evaluations. Computing capacity can also be used for downstream analyses during times of lower system utilization.

Vaidio Vista: Context checking with vision-language models

With Vaidio Vista, Vaidio introduces an additional verification level. Existing computer vision models first detect a possible object, behaviour or event. A vision-language model can then evaluate the scene in context.

This two-step approach combines fast, specialised recognition with broader scene understanding. The aim is to prioritize relevant events more precisely, reduce false alarms and give control centres or operations teams more context for their decisions. The human assessment and the defined reaction process still remain essential components of a reliable alarm chain.

What does Vaidio 10.0 mean for existing camera systems?

Vaidio is designed to reuse existing cameras and video management systems. Version 10.0 continues this approach and extends the utility of the existing infrastructure in three directions:

• More scaling: Additional cameras and locations can be planned more economically with more efficient GPU usage.
• More usable video material: In addition to live streams, stored material can also be included in analysis and research processes.
• More context: VLM-powered verification can classify selected events more accurately before a team responds.

Before introduction, camera perspective, image quality, network, storage, VMS compatibility, data protection and required computing power should still be checked. A real pilot shows which configuration works reliably for the respective application.

Classification by Synaedge

For Synaedge, Vaidio 10.0 is particularly interesting where a lot of standardized video analytics, multiple locations or extensive research and verification processes come together. The new version not only expands individual analysis functions, but also improves the basis for controllable enterprise operations.

Which platform and form of operation makes sense depends on camera inventory, analyses, data sovereignty, integrations and operating model. Synaedge examines these requirements and, if necessary, connects Vaidio with video management, control centres, access systems, APIs and existing business processes. For a specific classification you can use your Discuss the Vaidio project with us.

Editorially classified: Synaedge AG · As of: August 27, 2026. Performance values ​​and availability of individual functions depend on configuration, hardware and licensing.