Elon Musk announced on August 11 on the X platform that xAI's Grok V7 base model was pre-trained last week. The biggest highlight of this version is its native multimodal capabilities, which can directly process video and audio bitstreams without conversion to understand the content.
This means that Grok V7 can not only understand video images but also perceive subtle changes in speech, accurately identifying emotions and tonal emphasis in expressions, thus achieving a deeper semantic understanding.
At the same time, Musk also announced that the Grok4 model is now freely available to all users. Free users can perform a limited number of queries per day, and if more usage permissions are needed, they need to pay for a subscription. This move aims to expand the user base of Grok, making it more accessible and widely used by the public.
The native multimodal capabilities of Grok V7 indicate significant improvements in video and audio processing, opening up more possibilities for future AI applications. The free availability of Grok4 also reflects that xAI is balancing technological innovation with market penetration through different strategies.