I switched to GPT-5.6 Sol in Codex because it actually closes the engineering loop. It reads the repo, traces bugs across code, config, databases and deployment logs, edits multiple files, runs checks, and keeps iterating until the fix works. It feels less like chatting with AI!
@thsottiaux I switched to GPT-5.6 Sol in Codex because it actually closes the engineering loop. It reads the repo, traces bugs across code, config, databases and deployment logs, edits multiple files, runs checks, and keeps iterating until the fix works. more like shipping with an engineer.