shared older model numbers here https://www.latent.space/p/ainews-frontiercode-benchmarking https://www.latent.space/p/ainews-frontiercode-benchmarking tldr th

shared older model numbers here https://www.latent.space/p/ainews-frontiercode-benchmarking https://www.latent.space/p/ainews-frontiercode-benchmarking tldr theres been broad progress despite your observed regressions