To convert the public version of artificial intelligence, the statistics of an inaccessible experimental version were used.
This weekend, Meta released two new Llama 4 artificial intelligence models -light Scout and Medium -format Maverick, which, according to the company, exceeded GPT -4O and Gemini 2.0 flash in a wide range of popular trials. But, like this, statistics from meta are misleading users, The Verge reports.
In a press release on the release of AI meta models, they emphasized their great results in the LMarena-Platform rating, where users compare II models to chat mode and vote for the best. According to the company, Maverick took place in the second place from the 1417-Abt-4O-rating of the Openai and just a few behind Gemini 2.5 Pro.
However, users noticed a note in the META documentation, stating that not a public model was tested, but an experimental version, which was specially -optimized for communicating the conversation. The company did not immediately report it, but it was only confirmed that the customized version was used-llama-4-maverick-03-26-experimental, created for a better chat impression.
In response, LMARENA’s leadership accused Meta for their expectations from suppliers of AI models, and have begun updating their policies to avoid such situations in the future. The company believes that when suppliers can provide specially focused versions of their models for testing, simultaneously releasing completely different for the public, ratings like LMarene, have become more meaningful as real performance indicators.
We will remind you, earlier it was reported that a new test for artificial intelligence was not a single AI model.
Cloudflare created AI to deceive bots collecting information for AI
News from CORRESPONDENT.NET On the telegram and whatsapp. Subscribe to our channels https://t.me/KorresPondentNet and WhatsApp
Source: korrespondent
I’m Liza Grey, an experienced news writer and author at the Buna Times. I specialize in writing about economic issues, with a focus on uncovering stories that have a positive impact on society. With over seven years of experience in the news industry, I am highly knowledgeable about current events and the ways in which they affect our daily lives.