Looks like this already flipper over at polymarket

@TeddyWon That is a different benchmark

The whole episode that started with Dylan Patel saying OpenAI hadn't successfully pretrained a model in over a year was part of some choreography wasn't it

https://x.com/0xashensoul/status/1997639370669539462

Even conditional on OA overtaking Google before EOY, I still think Google has a strong chance of ending the year ahead.

@jim ok that's a market

Why is OpenAI still at 21 %? Is there a big release coming this month? GPT-5.1 High 1443, Grok 4.1 Thinking 1453, Gemini 3 Pro 1486.

@Lilemont /jim/openai-releases-new-reasoning-model

@jim Thank you!

@clementdupOz just FYI, the reason I only bet a little, is because the description of this market specifies "without style control". When style control is off, 2.5 pro still wins. and 3.0 is likely to be a lot higher.

@AGucci yo, emmm

@clementdupOz they now provide a direct link to the "Without Style Control" setting. See here for the current ranking the market resolution will be based on:
https://lmarena.ai/leaderboard/text/overall-no-style-control

what if gpt5.1 leapfrogs gemini 3?

@Dulaman very unlikely to at lmarena without style control specifically

Is this mainly because Google seems to goodhart the Lmarena leaderboard?

@CampbellHutcheson they goodharted it with gemini 2.5 pro. tbh i could see them not doing that anymore and gemini 2.5 pro staying #1 even after gemini 3 is released

Interesting to see how far Meta has fallen from grace