Elon Musk Teases Launch of Upgraded AI Chatbot, Grok-1.5

Elon Musk has recently hinted at the imminent release of an upgraded version of his company’s artificial intelligence (AI) chatbot, Grok. The reveal came through a social media post on March 29th, where Musk shared that the enhanced AI chatbot, Grok-1.5, might be launched as early as next week.

The new version will initially be available to early testers and existing Grok users in the social media platform. Musk also shared insights into the future development of Grok 2, suggesting that it could exceed current AI standards in all aspects.

What is Grok-1.5?

Grok-1.5 is an advanced version of the Grok-1 AI model, featuring improved reasoning capabilities and a context length of 128,000 tokens. This version represents a significant enhancement in the performance of the chatbot, particularly in coding and math-related tasks. Despite these improvements, it still falls short when compared to Google’s Gemini Pro 1.5 and OpenAI’s GPT-4.

According to xAI, Grok-1.5 scored a 50.6% on the MATH benchmark and a 90% on the GSM8K benchmark, which covers a wide range of grade school to high school competition problems. Additionally, it scored 74.1% on the HumanEval benchmark, evaluating code generation and problem-solving abilities.

Grok-1.5 can process information from significantly longer documents and handle more complex prompts while maintaining its instruction-following capability as its context window expands. The firm attributes these advancements to a custom distributed training framework based on JAX, Rust, and Kubernetes, enabling the team to prototype ideas and train new architectures at scale with minimal effort.

Grok Goes Open-source

In a significant move, xAI announced earlier this month that it had open-sourced the base code of Grok-1. This decision is a response to a legal action initiated by Musk against OpenAI, which he once co-founded. Musk claimed that OpenAI had deviated from its original commitment to prioritize open-source model development.

xAI clarified that the released code is “the raw base model checkpoint from the Grok-1 pre-training phase, which concluded in October 2023.” The model is not fine-tuned for any specific application, such as dialogue. The company mentioned that the model is licensed under Apache License 2.0.

With the tease of an upgraded Grok-1.5 and the promise of a future Grok 2, the AI world awaits the next steps from Musk and his team at xAI.