The New xAI Model?

Today’s letter:

The New Grok 1.5 Model (4 min)

xAI Elon Musk’s new AI company associated with 𝕏(formally Twitter) is releasing a new model: Grok 1.5

The Big Picture:

• Open Source First: The announcement follows xAI’s recent open-sourcing of Grok-1

• Grok-1.5 features improved reasoning, particularly in coding and math-related tasks.

• It can process longer contexts  (the ability for the user to input more information) of up to 128,000 tokens(1 token = roughly ¾ of a word) promising better understanding and performance.

The April Open Source Release of V1:

Why It Matters:

• Grok-1.5's enhancements position xAI to compete with major players like OpenAI's ChatGPT and Google's Gemini AI.

• Improved performance metrics, especially in math and language capabilities, aim to bridge the gap with competitors like ChatGPT

• Handling longer contexts signifies progress in AI understanding and application, potentially expanding Grok's utility.

How Grok 1.5 Stacks up:

• The MMLU benchmark: is a set of 15,000 multiple-choice questions across 57 topics, to test the general knowledge of Models

  • Grok 1.5 scored 81.3%

• The MATH benchmark: consisting of 12,500 high school and undergraduate level problems, assesses the mathematical reasoning

  • Grok 1.5 scored 50.6%

• The GSM8K benchmark: is a dataset of 8,000 grade school math word problems, designed to measure mathematical reasoning and problem-solving abilities 

  • Grok 1.5 scored 90%

Read More Here: