xAI has officially launched Grok 4.20, introducing four reasoning modes: Auto, Fast, Expert, and Heavy. The Auto mode dynamically selects between Fast and Expert, with Fast focusing on speed and Expert on deep reasoning. Founder Elon Musk announced on X that major upgrades will continue weekly, inviting user feedback on the new release. Grok 4.20, which exited beta testing today, achieved a 78% non-hallucination rate in tests by Artificial Analysis but scored 48 on the Composite Intelligence Index, behind competitors Gemini 3.1 Pro and GPT-5.4. The API is priced at $2 per million tokens for input and $6 for output, supporting up to 2 million tokens of long context.