pull down to refresh

The ELO on LMArena overall is down to 1498 right now, but it's still beating Grok 4.1 (by a hair or 2.) It definitely beats gpt-5.1 and also Claude 4.5 on coding, according to users.
What worries me is that open models aren't competitive right now, Kimi/GLM/Qwen are barely in the top 20.