# 0.6.0

## Release Information

- OS & Framework support - Tested with the following configurations:

> 
> 
> - Non-Gen AI:
> 
> 
> 
> > 
> > 
> > - Frameworks: ONNX, TFLite, Pytorch
> >         - Host Platforms: Linux-x86\_64 (Ubuntu 22.04), Linux-aarch\_64 (Ubuntu 22.04), Windows-x86\_64 (10+), Windows-arm64 (10+)
> >         - Target Platforms: Linux-x86\_64 (Ubuntu 22.04), Linux-aarch\_64 (Ubuntu 22.04), Windows-arm64 (10+), Android-arm64, QNX
>     - Gen AI:
> 
> 
> 
> > 
> > 
> > - Frameworks: ONNX, GGUF
> >         - Host Platforms: Linux-x86\_64 (Ubuntu 22.04)
> >         - Target Platforms: Android-arm64, Linux-aarch64
- QAIRT SDK - Tested with version (2.44.0, 2.45.0(Default), 2.46.0)

Note

For QAIRT SDK, other supported SDK versions can be listed using `qairt-vm fetch --list`.

## Highlights

- Default QAIRT SDK updated to 2.45.0
- Minimum supported QAIRT SDK updated to 2.44.0
- arm-Linux wheel support added (Linux-aarch\_64 host platform)
- New features in QAIRT Dev Python APIs (compatible with QAIRT SDK 2.45.0+, unless noted otherwise)

> 
> 
> - Gen AI Builder Enhancements
> 
> 
> 
> > 
> > 
> > - [Preview] Qwen3.5 and Qwen3 MoE model support in Gen AI Builder (requires SDK 2.46.0+)
> >         - LEQv2 encoding support in Gen AI Builder and embedding table utilities
>     - ONNX Optimizer New Passes and APIs
> 
> 
> 
> > 
> > 
> > - Public API for sequence/context length manipulation (`change_seq_length`, `change_context_length`, `change_seq_and_context_length`)
> >         - MoE (Mixture-of-Experts) adaptation pass (`adapt_moe`) for dynamic weight selection
> >         - IO Shape Rewriter pass for rewriting model input/output shapes across the graph
> >         - Conv layout optimization pass for improved Swin-Transformer and NHWC model performance
> >         - Linear-to-Conv transformation pass (`linear_to_conv`)
> >         - Parallelization pass to convert serial graph operations into parallel execution
> >         - Broader model architecture support in MHA2SHA conversion (additional operator coverage)
>     - HTP Backend Configuration Enhancements
> 
> 
> 
> > 
> > 
> > - Compilation options API (`set_compilation_options`) and backend extensions file loading
> >         - `CompileConfig.from_backend_extensions`, `populate_from_backend_extensions`, and `to_backend_extensions` for roundtrip backend extensions configuration
>     - Genie Execution Improvements
> 
> 
> 
> > 
> > 
> > - OE-Linux remote execution support for Gen AI API
> >         - OE-Linux native support for Gen AI API execution (requires QAIRT SDK 2.46.0+)
> >         - Improved LoRA adapter path resolution supporting both filename-only and relative paths
>     - Transform API Updates
> 
> 
> 
> > 
> > 
> > - `qairt.transform` now supports `adapt_moe` as a transform pass for MoE model preparation
>     - Deterministic AR/CL Conversion
> 
> 
> 
> > 
> > 
> > - New deterministic algorithm for resizing autoregressive (AR) and context-length (CL) dimensions of large language models, replacing the legacy implementation
> >         - MoE-aware AR/CL conversion support

## Resolved Issues

> 
> 
> - Gen AI Builder LoRA workflow failure (reported in 0.5.0) has been fixed. LoRA adapter path resolution and configuration handling now work correctly without manual workarounds.
> - Added import-error guard for Gen AI APIs on WoS. Due to a pybind library loading issue, Gen AI functionality is now safely disabled on Windows on Snapdragon rather than causing failures.

## Known Issues

> 
> 
> - Gen AI Builder is temporarily unsupported on Windows-arm64. Use Linux-x86\_64 as the host platform for Gen AI workflows.
> - Models with batch size greater than 1
> 
> 
> 
> > 
> > 
> > - A discrepancy exists between the qnn-net-run CLI and the Python API regarding data preparation for batch sizes
> > greater than 1. While the CLI automatically groups individual tensor paths from an input list into a batch,
> > the Python API requires inputs to be pre-batched (concatenated into a single raw file) prior to execution

Last Published: Jul 08, 2026

[Previous Topic
Known Issues](https://docs.qualcomm.com/bundle/publicresource/80-87189-2/topics/0_5_0.md) [Next Topic
0.7.0](https://docs.qualcomm.com/bundle/publicresource/80-87189-2/topics/0_7_0.md)