Syntiant NDP120 Sets New Standard for Energy Efficiency in Latest MLPerf Tiny v1.3 Benchmark Suite
Demonstrates Industry-Leading Energy Efficiency Across Varying Operating Modes
Irvine, Calif., September 30, 2025 – Syntiant Corp., the recognized leader in low-power edge AI deployment, today announced that its NDP120 Neural Decision Processor™ led the new streaming wake word detection test in the latest MLCommons’ MLPerf™ Tiny v1.3 benchmark suite, achieving both the lowest energy per inference and highest throughput.
On-device profiling shows the streaming workload uses a meager ~2% duty cycle at 50 MHz, leaving substantial capacity to run additional neural networks concurrently, such as noise cancellation, beamforming or other sensor-fusion tasks.
“Results of the MLPerf Tiny v1.3 benchmark builds on our strong performances in prior rounds, including static tests that demonstrated both industry-leading energy efficiency and much lower latency than typical microcontrollers,” said Jeremy Holleman, chief scientist at Syntiant. “By delivering best-in-class performance across multiple operating points, the NDP120 enables developers to optimize for throughput and power consumption, unlocking new use cases in speech, audio and sensor-driven AI applications.”
The new streaming audio benchmark highlighted the NDP120’s ability to process a continuous audio input efficiently in a test that mimicked a realistic deployment. Compared with traditional microcontrollers, the NDP120 delivers 10–100x better energy efficiency, making it an ideal solution for always-on applications such as keyword spotting, smart sensors and low-power audio recognition.
Full results of the MLPerf Tiny v1.3 benchmark suite can be downloaded here.
About the NDP120 + Modeling Toolchain
Built using the Syntiant Core 2™ programmable deep learning architecture, the NDP120 is designed to natively run multiple deep neural networks on a variety of architectures, such as CNNs, RNNs and fully connected networks. It proved its versatility by taking on the new streaming benchmark. The next-generation Syntiant Core 3™, now sampling to customers, offers even higher capacity.
Syntiant’s toolchain provides a straightforward path from reference or customer models to deployment on NDP-class devices. It offers compatibility with common layers and flexible pre/post-processing on the integrated DSP with typical networks using only a fraction of on-chip resources.
About Syntiant
Founded in 2017 and headquartered in Irvine, Calif., Syntiant® is Making Edge AI a Reality™ by delivering highly efficient processor, sensor, and software solutions. With more than 100 million purpose-built silicon and ML models deployed, along with billions of MEMS microphones and sensors, Syntiant’s technology is powering edge AI applications for speech, audio, sensor and vision processing worldwide. From earbuds to automobiles, the company’s turnkey solutions enable advanced edge AI capabilities across diverse consumer and industrial use cases. More information on the company can be found by visiting www.syntiant.com or by following Syntiant on X @Syntiantcorp or LinkedIn.
About MLCommons
MLCommons is an open engineering consortium with a mission to make machine learning better for everyone through benchmarks and data. The MLPerf benchmarks are full system tests that stress machine learning models, software and hardware and optionally measure power usage. The open-source and peer-reviewed benchmark suites provide a level playing field for competition that drives innovation, performance and energy efficiency for the entire industry. For additional information on MLCommons and details on becoming a Member or Affiliate, visit mlcommons.org.
Contact:
George Medici/Natalie Mu PondelWilkinson gmedici@pondel.com/nmu@pondel.com 310.279.5980