Seems like it went from Thinking and Instant (using Standard/Extended effort) with selected models to a step closer to Codex? Now it has Intelligence with Low, Medium and High for the selected model. submitted by /u/onehedgeman [link] [comments]