What has your experience been with these models so far?
Tell me:
- What are you building or testing?
- Which model are you using?
- What does one model do noticeably better than the other?
- Where does the model still need the most improvement?
- Which harness are you using? Cursor, Claude Code, Codex, or something else?
If you can, include a screenshot of the result. The most helpful screenshots show the task or prompt, the model’s output, and any errors or unexpected behavior.
Even a quick note helps everyone learn what these models are genuinely good at and where they still struggle.