X’s (formerly Twitter) AI chatbot Grok is about to get an upgraded model dubbed Grok 1.5 very soon, as announced by its maker xAI in a recent blog post. The startup headed by Elon Musk has just revealed its latest generative AI model which will be released “in the coming days.”
According to the benchmark results and specifications published by xAI, Grok 1.5 is expected to be a significant upgrade over its predecessor Grok 1.0. xAI adds that Grok 1.5 is going to boast “improved reasoning” when it comes to coding and math-related tasks.
The model significantly outperformed Grok-1, more than doubling its score on the widely recognized mathematics benchmark, MATH. Additionally, it achieved a score that was over 10 percentage points higher on the HumanEval test, which assesses capabilities in generating programming language and solving problems.
However, it is worth mentioning that benchmark performance does not always translate to reliable real-world use as AI models that often perform well on esoteric tests such as graduate-level chemistry exam questions have trouble working on interactions with an average person.
But an upgrade that is always useful in daily use is an increased context window. Quite like OpenAI’s GPT 4 Turbo, Grok 1.5 is also able to process contexts of up to 128,000 tokens. This is equivalent to more than 300 pages of text in a single prompt. “Context window” refers to the amount of input an AI model can take in and process at the same time.
Models equipped with limited context windows often face challenges in retaining details from conversations that have just taken place. On the other hand, models designed with extended context windows not only overcome this limitation but also excel in comprehending the data sequence they encounter, offering an enhanced understanding as a notable perk.
xAI describes in its blog post: [Grok-1.5 can] utilize information from substantially longer documents. Furthermore, the model can handle longer and more complex prompts while still maintaining its instruction-following capability as its context window expands.”
The Grok chatbot is known for responding to questions with “a rebellious streak,” as Musk has described it, and also discusses topics that are off-limits to other AI models. It remains to be seen whether Grok 1.5 will bring any changes in this regard.
Grok-1.5 is set to be released to early testers on platform X, bringing with it a suite of new features. Musk has teased potential capabilities such as thread summarization and reply, along with content suggestions for posts. It remains to be seen whether these enhancements will be introduced in the upcoming update.
The news comes soon after xAI open-sourced its Grok 1 AI model following a lawsuit targeted at OpenAI.
