xAI unveils Grok-1.5, a superior language model version powering chatbots on X

Last week, Elon Musk's xAI recently unveiled Grok-1.5, an improved version of its original large language model (LLM). The upgraded model boasts superior reasoning and problem-solving abilities, rivaling both open and closed LLMs like OpenAI’s GPT-4 and Anthropic’s Claude 3 (Although it falls short in context window performance compared to Gemini 1.5 Pro). It should also be noted that Grok 1.5 is not open-source, unlike the one announced a few weeks ago.

Grok-1.5 is set to drive xAI’s ChatGPT competitor chatbot on the X platform, with Grok-2, its successor, currently under development. Musk has suggested that Grok-2 will surpass existing AI on all performance metrics, though no specific release date has been disclosed.

The new model, inspired by "The Hitchhiker’s Guide to the Galaxy," is designed to help with knowledge acquisition for everyone, regardless of their background or political beliefs. It has surpassed competitors like Llama-2-70B and GPT-3.5 in benchmarks such as GSM8K, HumanEval, and MMLU. The model has shown impressive test scores, including 50.6% on the MATH benchmark, 90% on GSM8K, and 74.1% on the HumanEval benchmark. Early testers and existing Grok chatbot users on the X platform will have access to Grok-1.5 from next week, with a wider rollout planned later.