Google has renamed most of the new voices for its Gemini-powered Assistant on Nest devices. The change comes after Google introduced three new voices in June as part of the Gemini Assistant update for Nest Mini and Nest Audio. Although many names have changed, the voices themselves remain largely the same.
Voice Name Changes
For users in the Public Preview program, six out of ten voice names received updates. The new names continue the botanical theme but appear more familiar. For example, “Ivy” was renamed “Violet,” which carries a more positive association. “Verbena” changed to “Magnolia,” a name likely more recognised by users. However, “Eucalyptus” replaced “Calathea,” a switch that may not affect many since “Eucalyptus” is a better-known term.
Also read: No humans needed! Watch this robot replace its own battery and get back to work in seconds
Descriptions of the voices mostly stayed the same, except for “Fern,” which now has a “Warm” tone rather than “Bright.” Users report that the voices sound very similar across both old and new sets. Google allows listeners to compare the voices to hear the differences themselves.
Recently, Google teased upcoming improvements to Google Assistant aimed at addressing reliability issues. These enhancements are expected to come with Gemini’s broader rollout, possibly marking the end of the “Google Assistant” brand. Initially, Google seemed to keep the Assistant brand for Home devices, but with Gemini coming to Google TV, the shift toward rebranding is clear.
The new Gemini-powered Assistant features are planned to launch this fall, following a period of public testing that began last year.
Deep Think Expansion in Gemini
In other news, Google introduced Deep Think, a tool available in the Gemini app for Google AI Ultra subscribers. This addition is part of Google’s advanced AI suite. The company has also granted select mathematicians access to the full Gemini 2.5 Deep Think model, which recently competed in the International Mathematical Olympiad (IMO).
Also read: Oppo F29 Pro Review: Polished performer that delivers where it matters
Google says the latest model incorporates feedback from early testers and reflects recent research progress. This version builds on the one presented at Google I/O earlier this year. The prior model took hours to solve complex math problems, but the new release is faster and better suited for daily use. Internal evaluations show the model now achieves Bronze-level performance on the 2025 IMO benchmark, a notable step after the earlier gold-medal standard at this year’s competition.
