# @voooooogel — 2026-04-10

♥4 ↻0 · https://x.com/voooooogel/status/2042697037960089764

@kromem2dot0 yeah, which is why i think a METR-like "lowest common denominator environment" benchmark is ~fine, as long as you don't overindex on the y values or model-to-model comparisons. the shape of the curve seems to be accurate

tags: author:voooooogel, kind:tweet, thread-context, year:2026