The company said the new toolkit can cut model development and parameter extraction times from weeks to hours, helping semiconductor teams accelerate PDK delivery and support design technology co‑optimisation (DTCO). The release forms part of Device Modelling MBP 2026.
The launch comes as the semiconductor industry adopts increasingly complex technologies, including gate‑all‑around (GAA) transistors, wide‑bandgap materials such as gallium nitride (GaN) and silicon carbide (SiC), and advanced integration approaches such as chiplets and three‑dimensional stacking. While these architectures offer performance gains, they also introduce new challenges for device modelling and parameter extraction.
Keysight noted that traditional compact modelling workflows rely heavily on physics‑based models and manual tuning, often requiring engineers to adjust hundreds of interdependent parameters across multiple operating conditions. This process can take several weeks and may still fall short of achieving optimal accuracy under tight development timelines.
According to the company, the new Machine Learning Toolkit addresses these challenges by combining neural network architectures with machine‑learning‑based optimisation. The toolkit includes an ML optimiser, automated extraction flows and supporting utilities, enabling parameter extraction to be reduced from more than 200 manual steps to fewer than 10 automated steps. This approach is intended to accelerate PDK delivery, automate DTCO workflows and shorten time‑to‑market.
Keysight said the toolkit enables global optimisation of more than 80 parameters in a single run, capturing secondary effects, temperature variations and dynamic behaviour across DC, RF and large‑signal domains. The automated workflow is integrated into the existing Device Modelling platform and supports Python‑based customisation. Workflows are designed to scale across multiple technologies, including FinFET, GAA, GaN, SiC and bipolar devices, allowing modelling approaches to be reused across different process nodes.
Commenting on the release, Nilesh Kamdar, general manager of Keysight EDA, said, “AI/ML is fundamentally transforming the traditional workflows and methodologies of compact modelling. With the new Machine Learning Toolkit, we empower our customers to deliver more predictive, higher‑quality models in significantly less time – accelerating PDK development and helping them keep pace with rapidly evolving semiconductor technologies.”
Alongside the toolkit, Keysight announced updates across several other device modelling products.
Device Modelling MQA 2026 introduces new rules related to ageing model quality assurance for OMI and MOSRA, while Device Modelling WaferPro 2025 adds a remote‑control capability to support low‑frequency noise testing.
The latest A‑LFNA 2026 release also introduces new low‑frequency noise stress test functionality, enabling a more streamlined transition from stress testing to noise measurement.