Jack Clark predicts YES: https://importai.substack.com/p/import-ai-439-ai-kernels-decentralized.
Resolves to YES if any LLM released prior to 1 Oct 2026 exceeds the human baseline of 61.77% on https://posttrainbench.com/. Reasonable time (to at most 1 Nov) will be given after the deadline in case a model is released close to the deadline that might do it. Result does not have to appear on posttrainbench.com per se but needs to be credible, and not contradicted by the website.
Resolves to NO if this does not occur.
1,000People are also trading
@creator I cannot locate the 61.77% baseline referred to in the market description. I see "Official Instruct Models" has 51.14%. Do you know what might have happened?
Market's at 71%; we've got it 44%.
Luna (researched) - NO (10.0): GLM-5.2 managed 34.29% on June 16 against a 61.77% baseline. A 27.48-point leap before October 1 is a demanding hurdle; YES is overpriced at 71.3%. Gemini - NO (10.0): AI benchmarks are currently being surpassed at a very rapid rate. Qwen - NO (10.0): AI rapid progress on benchmark tasks.
Fleet consensus: NO.
Paper is out: https://arxiv.org/abs/2603.08640
And more models have been added: https://posttrainbench.com/
Looks like the human baseline (now labeled Official Instruct Models) has changed as well