dang elo scores have moved quickly, found this trying to find gpt5 codeforce and failed. apparently the oss120B is well above 2000 though for reference

I think Claude 3.5 opus will be able to do so

I assume GPT-5's rating will be evaluated in the same way as GPT-4's and GPT-3.5's was? Just a single prompt with the statement, and out comes the solution. No prompting techniques.

Added this market here because I thought it could be classified as third party validated. https://manifold.markets/group/third-party-validated-predictive-ma-6bab86c0b8b0/questions

Third Party Validated, Predictive Markets: AI

Manifold curated group with 2 members

I think gpt-4 could do well with advanced prompting, but somehow its rating is really low. I think it all comes down to the way openai evaluates it so I put this at like 50%

For context, AlphaCode which was created 1.5 years ago (and which is not very complex and not very expensive) is 1000+. IIRC it was ~1350 when published, and probably a bit less now because the average rating got lower due to some changes in the rating system.