xAI
xAI apparently rolling Grok 1.5 for some testers
Artificial Intelligence firm xAI is releasing the Grok 1.5 large language model version to a handful of early-access beta testers, suggests new information.
Released in November, Grok is a text-to-text generative AI chatbot that enables users to ask questions and get answers. It is currently integrated into social media site X, formerly known as Twitter. However, It is available only with Premium or Premium+ subscriptions.
According to xAI, Grok is modeled after Hitchhiker’s Guide to the Galaxy, a comedy science fiction franchise by Douglas Adams. Its design philosophy is reflected in its two modes, Regular and Fun.
With this release, xAI aims to compete against OpenAI, the company behind ChatGPT. It is owned by the OpenAI’s cofounder Elon Musk. However, Musk alleges that OpenAI is no longer open source and controlled by Microsoft, given its sheer size of investment in the ChatGPT maker.
The xAI team includes former researchers and AI specialists from Google’s DeepMind, OpenAI, Google Research, Microsoft Research, Tesla, and the University of Toronto.
Grok 1.5
Last month, xAI introduced Grok 1.5 with improved performance and advanced reasoning. Its benchmark test shows that the model architecture has been improved compared to Grok 1.0 released last year. It also closes the gap with competitors including Mistral Large, Claude 2, Claude 3 Sonnet, Gemini Pro 1.5, and GPT-4.
Grok 1.5 brought the capability to process long contexts of up to 128,000 tokens within its context window. Increased memory enables the model to utilize information from longer documents than its predecessor.
X user @cb_doge recently posted a screenshot of Grok saying “Grok v1.5 can explain memes”. The conversation shows the capability to explain memes, a new feature driven by Grok 1.5 Vision. It suggests that xAI is sending this new version to a few testers.
xAI Grok 1.5 explaining meme (Image Credit: X/@cb_doge)
The ability to explain memes is added with Grok 1.5 vision. It is a first-generation multimodal model that can process visual information including documents diagrams, charts, screenshots, and photographs. For now, xAI has not announced a wide rollout of its new Grok version.
