💨 Abstract

Meta's experimental AI model, Llama 4 Maverick, was found to have been optimized for conversationality and scored high on a crowdsourced benchmark, LM Arena. However, upon releasing the unmodified version, it ranked lower than competitors like OpenAI's GPT-4o, Anthropic's Claude 3.5 Sonnet, and Google's Gemini 1.5

Courtesy: techcrunch.com

Summarized by Einstein Beta 🤖