SpaceXAI has released Grok Voice 3.0, a new speech-to-speech model designed to listen, reason, call tools, and speak. According to reporting by MarkTechPost in the AI Tools section, the model has achieved a score of 86.9, placing it first on the Artificial Analysis Speech-to-Speech Index. For developers utilizing the service, pricing remains set at $0.08 per minute, alongside unchanged endpoints and events compared to the previous version, Grok Voice 2.0.
Performance and Pricing Details
The release marks a notable benchmark achievement for SpaceXAI in the conversational voice agent space. By securing the top position on the Artificial Analysis index, Grok Voice 3.0 outperforms competing conversational audio systems currently tracked on the benchmark. Developers can transition to the new version without incurring price changes or restructuring API endpoints, as the cost structure stays identical to Grok Voice 2.0.
Unanswered Questions and Technical Scope
Despite the release and benchmark results, several technical details remain unaddressed. The announcement did not specify the underlying parameter count, the specific training dataset composition, or the exact latency metrics experienced during tool calls. Furthermore, the timeline for broader platform rollouts and the extent of the model’s multi-turn reasoning improvements outside of benchmark tests are currently unknown.
As developers begin testing the new endpoints, further real-world performance data will clarify how Grok Voice 3.0 handles complex production workloads.
Lead image: Nicolas Foster / Pexels
