The new version of Elon Musk's Grok AI has improved coding and maths capabilities

xAI

Elon Musk-owned xAI has released an updated version of its large language model Grok-1.5 to early testers. According to the company, the new iteration demonstrates improved reasoning capabilities and skills, as well as the ability to handle longer contexts.

Here's What We Know

According to xAI, Grok-1.5 outperforms competing models such as OpenAI's GPT-4, Google's Gemini Pro 1.5 and Anthropic's Claude 3 Opus in a number of benchmarks related to mathematical tasks and coding.

In particular, Grok-1.5's results in MATH, GSM8K (maths problems) and HumanEval (encoding) benchmarks are significantly improved over the previous version and outperformed the competitors, with the exception of Claude 3 Opus in the HumanEval test.

Comparison of Grok-1.5 with other language models in popular benchmarks

In addition, the new version of Grok can handle contexts up to 128,000 tokens in length, which is 16 times the previous figure. This allows the model to combine information from larger sources for better situational awareness.

Despite the claimed progress in maths and coding, xAI did not provide details of Grok's improvements in other areas where it may still lag behind competitors.

Grok is currently only available to premium users of the X (formerly Twitter) platform owned by Ilon Musk. However, the chatbot is expected to be made available to regular X users in the near future.

The release of the updated version of Grok comes amidst the anticipation of OpenAI's next major update to ChatGPT, which promises to offer more human-like interaction with artificial intelligence.

Source: Engadget