Graiphic proudly participated in the GLA Summit held on June 23 and 24, 2025, where CTO Youssef Menjour introduced the LabVIEW community to groundbreaking innovations in AI graph technology: SOTA and ONNX GO.
Redefining the Role of ONNX in Deep Learning
During his presentation, titled “SOTA: Modern Deep Learning is LabVIEW Dataflow,” Youssef challenged conventional thinking about the ONNX ecosystem. Traditionally, ONNX has been treated as a static endpoint, used only as a format for trained models. At Graiphic, we see ONNX as something entirely different: a dynamic, editable starting point for designing, training, orchestrating, and deploying advanced AI systems.
Introducing SOTA, the First Fully Native ONNX Framework
SOTA (State Of The Art) is a visual deep learning framework built exclusively on ONNX and ONNX Runtime. It enables developers and engineers to design, train, and run AI models directly inside LabVIEW, without needing to rely on Python, Docker, or any external conversion tools.
What Makes SOTA Unique
-
100% native ONNX framework with no wrappers or custom scripts
-
Full support for both inference and training workflows
-
Visual and modular graph editing inside LabVIEW
-
Support for supervised learning, reinforcement learning, and custom architectures
-
Native orchestration engine capable of dynamically injecting ONNX subgraphs (e.g., loss functions)
-
Completely independent of VIPM with a custom installer
-
Made for engineers, researchers, and students alike
With SOTA, ONNX is no longer the end of the pipeline. It becomes the foundation of a new way to build AI, where models are not frozen assets but dynamic, reconfigurable graphs that evolve visually with your project.
ONNX GO: Going Further in Graph Orchestration
Alongside SOTA, Graiphic presented ONNX GO, a solution for orchestrating ONNX graphs in real time within LabVIEW. ONNX GO enables seamless deployment of dynamic models across heterogeneous hardware using ONNX Runtime and execution providers, all without a single line of code.
What Comes Next: ONNX GO HW
Looking ahead, Graiphic is preparing the next evolution: ONNX GO HW. This initiative will introduce standardized hardware nodes such as DMA, GPIO, memory buffers, and timers. It will allow ONNX graphs to interact natively with hardware-level operations, targeting System-on-Chip (SoC) platforms and enabling fully unified pipelines that combine AI, general processing, and hardware I/O.
Watch the Full Presentation
You can replay the full GLA Summit session and discover SOTA in action by watching the video here:
Download SOTA Today
SOTA is now available for download at https://graiphic.io/download/
Free licenses are available for students.
