# 0.10.0

## Release Information

- OS support - Tested with Windows-x86 (10), Windows-arm64 (10), Mac-arm64 (14), Linux (Ubuntu 22.04)
- QAIRT SDK - Tested with versions (2.44.0, 2.45.0)
- Model Visualization format support - ONNX, TensorFlow, PyTorch (1.x, 2.x), TFLite, Executorch Program, DLC, EAIX, HTP, and QNN JSON

## Highlights

- Graph Query

    - A new Graph Query panel allows users to define rules to match ops, static, or non-static tensors using regex-based or attribute-based conditions against node or tensor properties
    - Matched elements can be traversed directly in the graph using a match navigator, making it fast to locate and inspect specific nodes or tensors
    - Preset rule sets are available, covering tensor precision for both static and non-static tensors and EAIX-specific properties such as offloaded ops and cached tensors
    - Each rule can also specify visual styling for matched elements, enabling customization of node borders, background colors, and edge line styles and colors
- Model diff - Diff Customization Options and UI Flow Improvements

    - New Diff Customization options allow for fine-grained control over what is compared during a model diff. Includes the ability to ignore comparisons for op names, op attributes, tensor data, tensor shape, tensor names, and quantization encoding
    - Relative and absolute tolerance thresholds for tensor data comparison can be configured directly from the toolbar
    - Diff reports can now be re-run with different customization options, and reference and comparison file paths can be changed
    - All diff customization options are also available via Visualizer’s UI, Python API’s `diff()` module, and `qairt-visualizer diff` commands
- Composite Subgraph (Function) Support

    - Graph visualization now supports function-type nodes, enabling visualization of Mixture of Experts (MoE) models with components such as routing logic and multiple expert subgraphs
- QNN JSON Model Support

    - Adds support for visualizing models in the QNN JSON format, including QNN JSON model diffing
- UI enhancements

    - Shortcuts Panel: Each panel now includes a keyboard shortcuts button that opens a panel-specific overlay listing all available shortcuts for that panel

## Known Issues

- Diffing larger models in the QAIRT Visualizer Graph Panel (i.e., with 40K nodes or more or larger than 1.5 GB) may result in performance degradation or may fail to load altogether due to memory constraints. Upcoming releases will address these limitations to allow extra-large models to be loaded.
- Visualizing a model with performance and accuracy reports then clicking on any bar in the accuracy report causes all the panel views of the aggregate workspace to be updated, rather than just the corresponding tensor in the graph panel view
- DLC’s converted with QAIRT SDK 2.45 may fail to load op mappings.

    - Workaround: Convert ONNX models to DLC with QAIRT SDK 2.44 or lower

## Notices

- QAIRT Visualizer content may include China design and development contributions

Last Published: Jun 03, 2026

[Previous Topic
Known Issues](https://docs.qualcomm.com/bundle/publicresource/80-87189-1/topics/0_9_0.md) [Next Topic
0.10.1](https://docs.qualcomm.com/bundle/publicresource/80-87189-1/topics/0_10_1.md)