Elon Musk’s xAI has unveiled an upgraded version of its large language model (LLM) called Grok-1.5, following the recent open-sourcing of Grok-1. This new iteration promises enhanced reasoning and problem-solving capabilities, aiming to rival established LLMs like OpenAI’s GPT-4 and Anthropic’s Claude 3. While Grok-1.5 falls short of Gemini 1.5 Pro’s context processing abilities, it boasts improved performance across various benchmarks, including coding and math tasks.
Grok-1.5 achieved notable scores on benchmarks such as MATH (50.6%), GSM8K (90%), and HumanEval (74.1%). It also surpasses Grok-1 on the MMLU benchmark with a score of 81.3%, thanks to its expanded context window of up to 128,000 tokens. This upgrade enables Grok-1.5 to process longer prompts and documents more efficiently.
Despite its advancements, Grok-1.5 still trails behind Gemini 1.5 Pro, GPT-4, and Claude 3 on certain benchmarks. However, industry experts anticipate that Grok-2, currently in training, will surpass existing AI models across all metrics upon release. Brian Roemmele, a tech consultant, predicts that Grok-2 will be one of the most powerful LLM AI platforms available.
Grok-1.5’s deployment is scheduled to begin next week, initially targeting early testers and users of the Grok chatbot on the X platform. The rollout will be gradual, with xAI planning to introduce new features and improvements over time. Musk’s strategy of integrating Grok with the X platform aims to increase adoption, with recent updates making the chatbot available to a wider range of subscribers, including those with verified follower statuses.
