GPT-5.2-high LMArena scores released, OpenAI falls from #6 to #13

Posted by reed1234 5 hours ago

103Original

Comments

Comment by _nub3 3 hours ago

Cookie Banner conflicts with cloudflare anti bot stuff

Site is unusable.

Comment by reed1234 5 hours ago

While GPT-5.2 scores well on benchmarks, human preference is important for OpenAI’s consumer focused products.

Comment by aeonfox 3 hours ago

Arena Overview section is heavily biased towards languages. grok-4.1-thinking is worse than claude-opus-4-5-20251101-thinking-32k on every non-language metric by a large margin but somehow ranks higher overall, maybe because opus is way worse Spanish and Korean?