Google has released TranslateGemma, a set of open translation models based on the Gemma 3 architecture, offering 4B, 12B, and ...
NVIDIA PersonaPlex runs real time with listener signals and low overlap delay, helping callers feel heard while you resolve tasks more quickly.