\
  The most prestigious law school admissions discussion board in the world.
BackRefresh Options Favorite

OpenAI BENDS THE KNEE to Hegseth - link

https://x.com/FirstSquawk/status/2027580855208841652
ruby jap
  02/27/26
Mass consumer switches from ChatGPT to Claude inc.
ruby jap
  02/27/26
180. guess I'm sticking with gpt then fuck claude
Trip Parlor
  02/27/26
Overnight GPT and Claude will become political proxies. They...
ruby jap
  02/27/26
Codex 5.3 is clearly ahead right now though and gpt 5.3 will...
Trip Parlor
  02/27/26
I gave both Claude and GPT pay versions identical legal rese...
ruby jap
  02/27/26


Poast new message in this thread



Reply Favorite

Date: February 27th, 2026 10:07 PM
Author: ruby jap

https://x.com/FirstSquawk/status/2027580855208841652

(http://www.autoadmit.com/thread.php?thread_id=5839190&forum_id=2).#49701326)



Reply Favorite

Date: February 27th, 2026 10:08 PM
Author: ruby jap

Mass consumer switches from ChatGPT to Claude inc.

(http://www.autoadmit.com/thread.php?thread_id=5839190&forum_id=2).#49701329)



Reply Favorite

Date: February 27th, 2026 10:10 PM
Author: Trip Parlor

180. guess I'm sticking with gpt then fuck claude

(http://www.autoadmit.com/thread.php?thread_id=5839190&forum_id=2).#49701332)



Reply Favorite

Date: February 27th, 2026 10:16 PM
Author: ruby jap

Overnight GPT and Claude will become political proxies. They are very similar in capabilities as best I can tell.

(http://www.autoadmit.com/thread.php?thread_id=5839190&forum_id=2).#49701341)



Reply Favorite

Date: February 27th, 2026 10:17 PM
Author: Trip Parlor

Codex 5.3 is clearly ahead right now though and gpt 5.3 will be too once they release it. What's weird is that gpt is more censored right now though, which is annoying and surprising

(http://www.autoadmit.com/thread.php?thread_id=5839190&forum_id=2).#49701345)



Reply Favorite

Date: February 27th, 2026 10:21 PM
Author: ruby jap

I gave both Claude and GPT pay versions identical legal research questions. Both did great overall, but on one question ChatGPT told me correctly that my circuit hadn't ruled on the issue, while Claude didn't outright hallucinate a case but found a kinda similar holding and tried to oversell the extent to which it was what I was looking for.

So GPT won my personal test.

(http://www.autoadmit.com/thread.php?thread_id=5839190&forum_id=2).#49701350)