LMArena has updated the ranking of Meta's newly released open-source large model Llama-32-Maverick, from the previous 0th place to the 0th place!

This also really hammered the previous developers' doubts about Meta's Llama 4 model that provided a "special version" to LMArena for the ranking of the rankings.

4月6日，Meta發佈了最新的大模型Llama 4，包含Scout、Maverick和Behemoth三個版本。其中，Llama-4-Maverick在LMArena公佈的Chatbot Arena LLM 排行榜中的排名第二，僅次於Gemini 2.5 Pro。

However, as the effects of developers actually using the open source version of the Llama 4 large model have been exposed, Llama 0's reputation has plummeted. Some developers found that the version of Llama 0 provided by Meta to LMArena was different from the open-source version submitted to the community, so Meta was suspected of cheating.

On 8/0, Chatbot Arena officially issued a document confirming the above doubts from users, publicly stating that Meta provided them with a "special version" and considered updating the leaderboard.