xAI Grok4.20 Officially Launched with Four Inference Modes; Musk Says Major Updates Weekly

xAI's Grok4.20 has officially launched after ending its Beta test, introducing four inference modes: Auto, Fast, Expert, and Heavy. The Auto mode switches between Fast and Expert automatically, Fast focuses on quick responses, while Expert emphasizes deep thinking.

The Beta version of Grok4.20 set a record with a 78% non-hallucination rate in tests, but its comprehensive intelligence index was 48 points, lagging behind Gemini3.1 Pro and GPT5.4's 57 points. Its API pricing is $2 per million input tokens and $6 per million output tokens, supporting an ultra-long context of 200,000 tokens.

Elon Musk, founder of xAI, posted on X to collect users' first impressions of Grok4.20 and stated that major updates will continue to be pushed weekly.