Today, we are excited to launch the Grok Voice Agent API, which helps developers build voice agents that support dozens of languages, call upon various tools, and search data in real time. Grok Voice Agent is the fastest and most intelligent voice agent on the market. The Grok Voice Agent API is ranked #1 on Big Bench Audio, the leading audio inference benchmark. Grok is nearly 5x faster than its closest competitor. @ArtificialAnlys The Grok Voice Agent API leads the industry in cost-effectiveness. Developers pay a flat fee of only $0.05 per minute. We built our entire voice technology stack ourselves, training our VAD, word segmenter, and audio model from scratch. In blind tests with OpenAI, Grok consistently ranked as the best model across all aspects, including pronunciation, accent, and prosody. Tesla is a key design partner for the Grok Voice Agent API, which currently powers Grok in millions of vehicles. Grok is like a natural extension of your Tesla, thanks to its dedicated tools for accessing vehicle status, planning routes, and controlling navigation. The Grok voice agent can perform tasks and find information instantly. Developers can easily integrate custom tools or leverage xAI's real-time search capabilities on X and the web. We are excited to offer a variety of expressive voices for the Grok voice agent API, including Ani, Eve, and Leo. Our voices sound natural and fluent in everyday conversation and excel in pronouncing specialized terminology in fields such as healthcare, finance, and law. The Grok voice agent API is compatible with the OpenAI Realtime API specification and is also available through the official xAI LiveKit plugin. In the coming weeks, we will be releasing standalone text-to-speech and speech-to-text interfaces, along with more powerful audio models. To get started, try our voice testing platform, where you can test various voices directly in your browser. We can't wait to see your work!