SpaceXSI introduced Grok Voice 3.0 in an Oct. 14, 2026, post. It described the speech-to-speech model as responding in under a second and excelling at resolving customer calls. The post also claimed it costs less than other frontier models, without giving a cost comparison.
A day earlier, Fish Audio introduced Drama 3 as an open-weight text-to-speech model. Fish Audio says users can control how a single word sounds, change its tone and shift voice mid-sentence. In a separate Oct. 14 post, GenAI_is_real said SGLang Omni supports Drama 3 INT4. The poster said a quantized checkpoint reduced engine-weight memory from 37.0 GB to 20.8 GB, with comparable voice-cloning quality in their Seed-TTS EN evaluation. They said a Hugging Face Space would soon run the model on SGLang Omni.
One customer-service use of xAI voice was still being tested. In an Oct. 19 post, developer damian_soong described a phone line using xAI voice, a Twilio toll-free number, their own server and Richpanel. They said the agent can check orders, send tracking links and answer product questions, but cannot yet change customer accounts. The line was in the final stages of testing, they wrote, and they hoped it would reduce tickets for simple support questions.
