Grok Voice 3.0 is reaching existing API users through an automatic model switch. The October 14 launch post says agents using grok-voice-latest have already moved to the new version without changes, at the same $0.08-per-minute price as Grok Voice 2.0.
The company describes customer support as a major deployment. It says Starlink call centers use Grok Voice for 20,000 calls a day across multiple languages, with roughly 70% resolved without a human in the loop. The launch post says an agent built on Grok Voice 3.0 can verify customers, troubleshoot hardware, issue credits and ship replacement equipment.
A leading index score, with mixed component rankings
Artificial Analysis’s own evaluation puts Grok Voice 3.0 first on its Speech to Speech Index at 86.9, up from 81.3 for Grok Voice 2.0. The index averages speech reasoning, agentic performance, user preference and task success. The organization tested the Rex voice through the Realtime API.
That overall lead does not mean it won every component. On , Artificial Analysis’s benchmark for agentic performance, Grok scored 71.6%, behind GPT-Live-1 (Astra, medium) at 74.5%. In the , it ranked third in preference but first in task success, at 94.8%.
