Elon Musk has announced that the Grok-2 AI model from xAI will be launched in August. This upcoming large language model (LLM) is anticipated to bring new features and enhanced performance compared to its predecessor. Musk also revealed plans to decrease the reliance on Internet-sourced training data to improve response quality. Details about the new AI model remain limited. Previously, in April, xAI introduced the Grok-1.5 Vision model, which processes images and responds to related queries.
The announcement follows xAI’s initial preview of the Grok-1 model in November 2023. Developed in just two months, Grok-1 became known for its witty and sarcastic responses. Unlike other LLMs like ChatGPT, Gemini, and Claude, Grok-1 has not been made publicly available.
In a recent post on X (formerly Twitter), Musk confirmed the August release of Grok-2. His comment was in response to a video interview with Cohere CEO Aidan Gomez, discussing the issue of chatbots exhibiting similar sentence structures due to being trained on overlapping data sets. Musk acknowledged this problem, noting that significant effort is required to remove Internet training data, which contributes to the homogeneity of chatbot responses. He assured that Grok-2 would be a “giant improvement” in this regard.
While Musk did not provide further specifics about Grok-2, he highlighted the challenges of reducing reliance on Internet data for training AI models. The Grok-1.5 Vision model, previewed in April, incorporates computer vision to handle visual media inputs, with a context window of 1,28,000 tokens, though it is not publicly available.
The current Grok AI chatbot can be accessed via the X Premium subscription, which costs Rs. 566.67 per month, and is also included in the X Premium+ subscription priced at Rs. 1,133.33 per month.