Watched the DevDay stuff yesterday and for the first time I'm actually rethinking my $200 Claude Max sub.
retained text
Watched DevDay yesterday and for the first time I'm seriously thinking about switching. GPT-6.1 Sol scores 52 on Artificial Analysis vs 58 for Opus 5.5, so Claude's still smarter. But it's roughly $0.72 per task vs $5.98. Most of my day is Laravel refactors, python scripts and tests. I don't need the smartest model alive, I need one that doesn't burn through my limits by Wednesday. Throw in Codex cloud execution, PR review, Dots and the new Pro tier... they're clearly coming for the Claude Code crowd. What's holding me back is the move. Months of tuned project instructions, knowledge files, memory. ChatGPT has no import and Claude's export is just a JSON dump. Rebuilding it all by hand sounds miserable. Probably going to run both for a month, move one project I use daily and see how Sol + Codex does on real work. Anyone already jumped? Does Sol hold up on a real codebase or fall apart at 50 files?
document fetched 2026-10-01 00:24 · canonical https://reddit.com/r/ClaudeAI/comments/1wuatc8/watched_the_devday_stuff_yesterday_and_for_the