Download

NPU Benchmark is currently in development. Release date is to be determined. Follow the project on GitHub for the latest updates.

🚧 In Development

Release Date: TBD

We are building and validating the NPUBench Standard v1.0 methodology across three platforms. When ready, the tool will be released as open source under Apache 2.0.

3 Platforms
Apache 2.0
Windows 11

System Requirements

From NPUBench Standard v1.0

Qualcomm
Snapdragon X Series
OS Windows 11 ARM64
Min Driver ≥ 30.0.140.1000
Min Memory 32 GB
Execution Provider NativeQNN
AMD
Ryzen AI 300 Series (XDNA2+)
OS Windows 11 x64
Min Driver ≥ 32.0.203.280
Min Memory 32 GB
Execution Provider OrtGenAI-RyzenAI
Intel
Core Ultra Series 2
OS Windows 11 x64
Min Driver ≥ 32.0.100.4297
Min Memory 16 GB
Execution Provider NativeOpenVINO

Driver version requirements sourced from MLPerf Client v1.5 reference data.

Benchmark Models

From NPUBench Standard v1.0 Section 6

Model Quantization License
Llama 3.2 3B Instruct int4 Meta Llama 3 Community License
Llama 3.1 8B Instruct int4 Meta Llama 3 Community License
Phi-3.5 Mini Instruct int4 MIT License
⚠ Llama model results must not be used to train or improve non-Llama AI models. See Meta Llama 3 Community License Agreement.

Stay Updated

Follow the project on GitHub for development updates, or join the NPU GenAI developer community to discuss NPU inference and benchmarking.

GitHub ↗ Community ↗
⚖ Legal Notice & Disclaimer

· NPU Benchmark (NPUBench) is an independently developed benchmark standard. It has no affiliation with or endorsement by MLCommons Association.

· MLPerf® is a registered trademark of MLCommons Association. This tool's methodology references MLPerf Client v1.5 for metric definitions and quality thresholds, but does not constitute or claim official MLCommons certification.

· Use of Llama 3.x models (Meta Platforms) requires compliance with the Meta Llama 3 Community License Agreement. Test results generated using Llama models must not be used to improve non-Llama AI models. Commercial use with >700M monthly active users requires a separate license from Meta.

· Phi-3.5 Mini Instruct (Microsoft) is used under the MIT License.

· NPUBench source code is released under the Apache 2.0 License.

· Test results generated by this tool do not represent the official position or endorsement of any hardware vendor (Qualcomm, AMD, Intel).

· All benchmark scores are measured under specific hardware and software conditions. Results may vary depending on system configuration, driver version, and operating environment.