SpaceXSI debuts Grok Voice 3.0, saying it tops Artificial Analysis' speech-to-speech index, responds in under a second, and costs less than rivals, at $0.08/min

Speech-to-speech AI models now respond in under a second and cost $0.08 per minute.

SpaceXSI has launched Grok Voice 3.0, a new speech-to-speech AI model. The company claims it is the most intelligent voice model available, achieving the top ranking on Artificial Analysis's independent Speech-to-Speech Index with a score of 86.9. It also demonstrates superior performance in conversational dynamics, scoring 97% on the Full Duplex Bench. Grok Voice 3.0 offers class-leading latency, responding in under a second, and is priced competitively at $0.08 per minute, the same as its predecessor. The model is already in use across Tesla and SpaceX services, handling millions of customer interactions daily, with a reported 70% resolution rate without human intervention in Starlink call centers. SpaceXSI asserts that Grok Voice 3.0 also outperforms rivals in cost-efficiency for task completion.

The release of Grok Voice 3.0 signifies a significant advancement in AI's ability to handle natural, real-time voice conversations, particularly in customer support scenarios. Its leading scores on comprehensive benchmarks for accuracy, reasoning, and conversational flow suggest a new standard for voice agent performance. The model's low latency and competitive pricing position it to disrupt the market for enterprise voice solutions, potentially lowering costs and increasing efficiency for businesses relying on customer service AI.

This development is likely to intensify competition among AI providers in the speech-to-speech domain, pushing rivals to accelerate their own performance and cost optimizations. Businesses evaluating AI solutions for customer interaction should now consider models that not only understand and respond but do so with human-like speed and minimal delay. Future developments will likely focus on further reducing latency, improving nuanced understanding, and expanding multimodal capabilities.