According to foreign media reports, Elon Musk's artificial intelligence company xAI announced the launch of the Grok Voice 3.0 voice model, now available via the Voice Agent API, priced at US$0.08 per minute (approximately HK$1), the same as the previous generation. xAI stated that Grok Voice has been deployed in Tesla vehicle voice interactions and at SpaceX's Starlink customer service center, which handles approximately 20,000 multilingual customer calls daily, with about 70% resolved without human intervention.
According to test results released by independent AI evaluation platform Artificial Analysis, Grok Voice 3.0 scored 86.9 on the Speech-to-Speech Index, ranking first in its debut evaluation. For comparison, OpenAI's GPT-Live-1 (Astra, medium reasoning configuration) scored 83.2, Google Gemini 3.8 Live Extended Thinking (high reasoning configuration) scored 82.6, and GPT-Live-1 (Sol, low reasoning configuration) scored 80.1.
Grok Voice 3.0 improved by 5.6 points over the previous generation Grok Voice 2.0's 81.3, and by 13.2 points over Grok Voice 1.0's 73.7.
Grok Voice 3.0 shows significant improvement in voice agent performance
On specific metrics, Grok Voice 3.0's performance shows divergence. In task success rate, the model ranked first with 94.8% in the Speech Agent Arena test, slightly above Grok Voice 2.0's 94.6%. In voice agent execution capability, it scored 71.6% in the Tau Voice test, second only to GPT-Live-1 (Astra, medium configuration) at 74.5%, and significantly higher than Grok Voice 2.0's 56.5%.
In audio reasoning, it scored 98.5% in the Big Bench Audio test, surpassing Google's related model at 97.7% and the previous generation Grok Voice 2.0 at 97.2%. However, in user preference, its Elo rating is 1070, lower than Gemini 3.1 Flash Live Minimal's 1100 and Gemini 3.8 Live's 1088, ranking third.
For voice agents requiring real-time interaction, response latency is a key metric affecting user experience. Artificial Analysis's testing shows that in the Big Bench Audio test, Grok Voice 3.0's average first audio response time is 0.81 seconds, slightly slower than Grok Voice 2.0's 0.72 seconds but still under 1 second. xAI stated in its official release that Grok Voice 3.0 can respond to customers in under 1 second at a cost lower than industry-leading models.