Hey @OpenAI@ChatGPTapp@thsottiaux any ETA on subagent model layering issue fix for Codex app? Its in the latest cli alpha, but not codex app. With the new models, this feels essential to avoid burning through the 5-hour usage window so quickly.
I had a similar issue so switch to this path and it has been working really well. Based it on all the recent benchmarks.
main: 5.6 gpt luna max
subagents:
- [oracle]: gpt 5.6 sol xhigh
(planning, debugging, hard problems)
- [librarian]: gpt 5.6 terra high
(internal researcher)
- [worker]: gpt 5.6 luna medium
(tool called, edits, reads, etc....)
Just make sure you update your global AGENTS.md to encourage their usage.
@trashh_dev No worries man. It's a learning experience. Sometimes it takes a few tries to get over the adrenaline dump and learn to relax and fight your game. Just remember most people are scared to even show up. So you're already beating them π€π½
One UI update Iβd love to see in the @cybertruck :
Allow us to select the viewing content for the rear screen from the front while driving ππ½
Itβs the same as searching and selecting a song/playlist.
@siddawa
@elonmusk any ETA on FSD (supervised) on CyberTruck. I'm happy to Beta test π€π½ @Apple watch car key would be rad too please. #askingforafriend π¬
If I could rid the programming world of one thing it would be `else` clauses. For the rare times it's the right choice, I would allow you to ask me for permission. π