Just used Opus 4.6 via GitHub Copilot. It feels very different. Inference seems slow for now. I guess Opus 4.6 has adaptive thinking activated by default.
> "We have high thinking as default + adaptive thinking, first time we’ve run with these settings..."