Techmeme

LMArena says it's updating its leaderboard rules after Meta secured the #2 spot with a Llama 4 Maverick version it noted in fine print was not released publicly (Kylie Robison/The Verge)

Kylie Robison / The Verge: LMArena says it's updating its leaderboard rules after Meta secured the #2 spot with a Llama 4 Maverick version it noted in fine print was not released publicly  —  With Llama 4, Meta fudged benchmarks to appear as though its new AI model is better than the competition.

Apr 8, 2025 - 04:19
 0
LMArena says it's updating its leaderboard rules after Meta secured the #2 spot with a Llama 4 Maverick version it noted in fine print was not released publicly (Kylie Robison/The Verge)

Kylie Robison / The Verge:
LMArena says it's updating its leaderboard rules after Meta secured the #2 spot with a Llama 4 Maverick version it noted in fine print was not released publicly  —  With Llama 4, Meta fudged benchmarks to appear as though its new AI model is better than the competition.

Read More

Tags:

Previous Article

Nintendo Won't Reveal Donkey Kong's Switch 2 Dev, But We Might Already Know

Next Article

How to enable and use RCS for secure cross-platform messaging

Related Posts

Nintendo delays preorders for the Switch 2 from April 9, 2025 due to concerns about Trump's tariffs but the June 5 launch date remains unchanged (Emma Roth/The Verge)

Nintendo delays preorders for the Switch 2 from April 9...

Apr 4, 2025  0

Filing: the man Rippling has accused of feeding stolen customer and employee info to Deel is Keith O'Brien, a global payroll-compliance manager in Dublin (Rolfe Winkler/Wall Street Journal)

Filing: the man Rippling has accused of feeding stolen ...

Mar 18, 2025  0

A survey of attendees at the WSJ's CIO Network Summit: 61% are experimenting with AI agents, but 21% are not using them at all, citing reliability concerns (Belle Lin/Wall Street Journal)

A survey of attendees at the WSJ's CIO Network Summit: ...

Feb 12, 2025  0