Cadence reveals Maestro, fine-tuned AI models built to run specialized chip design agents, with early access coming for 'select customers' in Q4 2026

Cadence has introduced Maestro, a new line of AI models specifically designed for its chip design agents, according to a press release prior to the company's earnings announcement. These models demonstrate up to a 30% improvement in quality and up to a 5x increase in token efficiency compared to general-purpose AI models. Early access to these models will be provided to select customers in Q4 2026.

Customers can utilize Maestro on Cadence's cloud, public clouds, or customer-controlled on-premises environments, and they can also fine-tune the models using their own proprietary data, designs, and workflows. This differs from Synopsys' GPT-Synopsys, which relies on OpenAI's hosting infrastructure and does not incorporate customer data into its model training.

Cadence's Maestro utilizes its AI Super Agents, which "interpret design intent, decompose tasks" and "orchestrate the Maestro portfolio with other leading frontier and open-weight models." The models also incorporate reinforcement learning, enabling the feedback loop of improved results, and can be run on Nvidia's Nemotron, among other models.