Meta’s benchmarks for its new AI models are a bit misleading

  • staffstaff
  • AI
  • April 6, 2025
  • 0 Comments

One of the new flagship AI models Meta released on Saturday, Maverick, ranks second on LM Arena, a test that has human raters compare the outputs of models and choose which they prefer. But it seems the version of Maverick that Meta deployed to LM Arena differs from the version that’s widely available to developers. […]

  • Related Posts

    Anthropic CEO wants to open the black box of AI models by 2027

    Anthropic CEO Dario Amodei published an essay Thursday highlighting how little researchers understand about the inner workings of the world’s leading AI models. To address that, Amodei set an ambitious goal for…

    Continue reading
    How do you define cheating in the age of AI?

    This AI startup raised $5.3 million to help people “cheat on everything.” But in the age of AI, how do you define cheating? Columbia University recently suspended student Roy Lee…

    Continue reading

    Leave a Reply

    Your email address will not be published. Required fields are marked *

    You Missed

    Anthropic CEO wants to open the black box of AI models by 2027

    • By staff
    • April 25, 2025
    • 2 views

    OpenAI rolls out a ‘lightweight’ version of its ChatGPT deep research tool

    • By staff
    • April 24, 2025
    • 2 views

    How do you define cheating in the age of AI?

    • By staff
    • April 24, 2025
    • 2 views

    Perplexity CEO says its browser will track everything users do online to sell ‘hyper personalized’ ads

    • By staff
    • April 24, 2025
    • 3 views