# Cadence launches Maestro, which it says offers up to 5x token efficiency for chip-design agents > Source: > Published: 2026-10-26 15:49:22+00:00 # Cadence launches Maestro, which it says offers up to 5x token efficiency for chip-design agents **Select customers are expected to get early access in Q4; Cadence has not named the comparison model or defined the 30% quality measure.** By [Ryan Merket](https://runtimewire.com/author/ryan-merket) · Published Primary source: [Tom's Hardware](https://www.tomshardware.com/tech-industry/semiconductors/cadence-reveals-maestro-fine-tuned-ai-models-built-to-run-specialized-chip-design-agents-with-early-access-coming-for-select-customers-in-q4-2026-claims-up-to-30-percent-higher-quality-results-and-up-to-5x-token-efficiency-over-a-general-purpose-ai-model) ## Why it matters Cadence is moving repeated chip-design agent work into specialized models and letting customers choose where those models run and what data they learn from. The cost and quality claims still need measurable customer results. Cadence [introduced Maestro on October 26th](https://www.cadence.com/zh_TW/home/company/newsroom/press-releases/pr/2026/cadence-introduces-maestro-a-domain-specialized-ai-model.html?ref=runtimewire), a portfolio of specialized AI models for its chip-design agents, and claimed initial results of up to [5x greater token efficiency](https://www.cadence.com/zh_TW/home/company/newsroom/press-releases/pr/2026/cadence-introduces-maestro-a-domain-specialized-ai-model.html?ref=runtimewire) than a general-purpose model.… When you purchase through links in our articles, [we may earn a small commission](https://runtimewire.com/affiliate-monetization-standards). This doesn’t affect our editorial independence.