Would be interesting to measure useful output vs compute time, as metric of how effective the models are.

Wild . Here my Proof of walking outside https://image.nostr.build/83664f8b0450dafece240d46df5bde4988e32d6d99c6ab1da724650461f55042.jpg

I only recently made the workflow update to cycling between 4 different projects where are agents are simultaneously working. Crazy times, even compared to ~4 months ago. What models do you use (TENEX, I assume you refer to here?)? Only the latest and greatest?

Just watched your pod with Preston, looking forward to reading your long form content.