xAI introduces Grok Voice Think Fast 2.0 with major accuracy and conversational gains

xAI has unveiled Grok Voice Think Fast 2.0, its latest voice AI model focused on delivering enhanced transcription accuracy and smarter conversational ability. This release introduces a significant leap over earlier iterations, as well as competing products. According to xAI, Grok Voice Think Fast 2.0 achieves 1.5 to 2.0 times better transcription accuracy than Deepgram Nova 3 and ElevenLabs Scribe v2, and is 1.4 times more accurate than its previous version across 24 evaluated languages and thousands of test phrases. The model’s accuracy advantage expands to nearly tenfold over leading speech-to-text systems in challenging settings with background noise or telephony compression. In addition to improved transcription, Grok Voice Think Fast 2.0 features parallel reasoning while speaking. This capability allows it to interpret complex queries and deliver intelligent responses with no increase in latency. Building on this, xAI applied intensive reinforcement learning so that the model emu...

Read Original

Related