did you address this https://github.com/openai/codex/issues/6426 ?

how much more token efficient is this compared to 5.0

had to use 5.0 because 5.1 was eating tokens like crazy and seemed like a slight incremental improvement barely noticeable