Elon Musk’s xAI has announced an upgraded version of its large language model (LLM) called Grok-1.5. This new version brings enhanced reasoning and problem-solving capabilities, closing in on the performance of other popular LLMs such as OpenAI’s GPT-4 and Anthropic’s Claude 3. While Grok-1.5 falls slightly behind Gemini 1.5 Pro in terms of context window size, it is still capable of processing long contexts. Musk stated that Grok-1.5 will power xAI’s ChatGPT-challenging chatbot on the X platform, with the successor model, Grok-2, currently in the training phase.
Grok-1.5 builds upon the success of its predecessor, Grok-1, which had outperformed other models on benchmarks such as GSM8K, HumanEval, and MMLU. The new model achieves significant improvements across all major benchmarks, including coding and math-related tasks. It scored 50.6% on the MATH benchmark, 90% on the GSM8K benchmark, and 74.1% on the HumanEval benchmark. On the MMLU benchmark, it scored 81.3%, outperforming Grok-1 by a significant margin.
One of the key features of Grok-1.5 is its context window of up to 128,000 tokens, allowing it to process large amounts of information in one go. This makes it more suitable for analyzing, summarizing, and extracting information from long documents. It can also handle longer and more complex prompts while still following instructions accurately.
Grok-1.5 not only outperforms its predecessor but also closes in on popular open and closed-source models such as Gemini 1.5 Pro, GPT-4, and Claude 3. It slightly falls behind these models on benchmarks such as MMLU and GSM8K, but it outperforms them on the HumanEval benchmark. Musk believes that Grok-2, the successor to Grok-1.5, will surpass current AI models on all metrics, making it one of the most powerful LLM AI platforms upon its release.
xAI plans to start deploying Grok-1.5 next week, initially making it available to early testers and existing Grok chatbot users on the X platform. The rollout will be phased, with the model continuously improving and introducing new features. xAI also plans to introduce a new unhinged fun mode for the chatbot. Musk’s decision to make Grok available on X was aimed at driving adoption for both Grok and the platform. Initially priced at $16 per month for Premium+ subscribers, Musk later expanded access to all Premium subscribers paying $8 per month. He also announced that followers with a certain level of verified subscriber followers will receive Premium and Premium+ subscription benefits, including Grok, for free.
Overall, Grok-1.5 represents a significant improvement in xAI’s language model capabilities and brings it closer to competing with other well-known LLMs in terms of performance. With the upcoming release of Grok-2, xAI aims to surpass current AI models and further solidify its position in the AI landscape.