Do Son OpenAI launches GPT-5.2 ("Red Alert") to reclaim the top spot from Gemini 3 Pro. The model achieves perfect math scores and promises unparalleled reliability for enterprise tasks.
LMArena
xAI released Grok 4.1 Thinking, which immediately topped the LMArena leaderboard (Elo 1483), pushing Gemini 2.5 Pro to #3. Grok 4.1 also cut hallucinations by 70%.