In a news update from Dongcha Beating AI, xAI has announced the release of its speech-to-speech model, Grok Voice 3.0, which has achieved a top score of 86.9 in the Speech-to-Speech Index of the Artificial Analysis independent test. This score surpasses Grok Voice 2.0's 81.3, GPT-Live-1's 83.2, and Gemini 3.8 Live's 82.6. The model boasts an end-to-end latency of under one second. It is now available through the Voice Agent API, priced at $0.08 per minute, the same as its predecessor, with automatic switching using the grok-voice-latest agent. Official sources state that Grok Voice handles millions of conversations daily at Tesla and SpaceX, with the Starlink customer service center receiving approximately 20,000 calls each day, about 70% of which are resolved without human intervention, and roughly a quarter of sales calls resulting in purchases before hang-up. The accuracy of ServiceNow EVA-Bench is reported at 74%, while τ-Voice stands at 71.6%, indicating performance comparable to GPT-Live-1 but with approximately 40% lower latency.
All Comments