Musk's innovative company, xAI, has introduced the Grok Voice 3.0 voice model. This cutting-edge model can now be accessed via the Voice Agent API at a competitive rate of $0.08 per minute, maintaining the same pricing structure as its predecessor. The Grok Voice 3.0 has already been integrated into voice interaction systems within Tesla vehicles and the customer service hub of SpaceX's Starlink. Impressively, Starlink's customer service center manages around 20,000 multilingual calls each day, with approximately 70% of these calls being resolved without any human assistance.
In a comprehensive voice-to-voice index test, Grok Voice 3.0 made a remarkable debut, securing the top spot with a score of 86.9. This achievement underscores significant enhancements across several core metrics when compared to its forerunner. The model excels in task success rates, surpassing Google's relevant models in audio reasoning capabilities. Furthermore, it ranks second only to the medium-configuration version of GPT-Live-1 in terms of voice agent execution prowess, and it holds the third position in user preference rankings.
Notably, Grok Voice 3.0 boasts an average first audio response time of just 0.81 seconds. Company officials have asserted that the model consistently responds in under one second, while also offering cost advantages over other industry-leading models.
