The most prestigious law school admissions discussion board in the world.
|
 |
 
|
ChatGPT Plus 5.2 is *almost* as good as Claude's Opus 4.5
Poast new message in this thread
Date: January 6th, 2026 11:18 PM Author: arousing claret shrine (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49568112) |
Date: January 7th, 2026 6:17 AM Author: Thirsty People Who Are Hurt Headpube 5.2 is significantly better IMO. In terms of getting sane and accurate answers. Claude is better at writing though. GPT still has a janky style to it’s outputs (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49568388) |
 |
Date: January 7th, 2026 6:23 AM Author: scarlet dashing marketing idea mad-dog skullcap Chatgpt writes
Like Jordan peterson
Withdrawing from benzos on
Twitter |
|
Date: January 7th, 2026 8:37 AM Author: Disturbing corn cake Your AI threading is the most retarded shit on xoxo rn (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49568494) |
|
Date: January 7th, 2026 9:40 AM Author: arousing claret shrine Explain? (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49568617) |
|
Date: January 7th, 2026 10:00 AM Author: Gay Stimulating Meetinghouse it still made up a case when i asked it a fairly specific question. Made up case name, cite, what it held, etc. (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49568677) |
|
Date: January 7th, 2026 10:25 AM Author: Curious bat-shit-crazy striped hyena church building you need to ground it with a claude skill and integrate exa search etc (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49568719) |
|
Date: January 7th, 2026 10:27 AM Author: arousing claret shrine (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49568722) |
|
Date: January 10th, 2026 2:15 AM Author: arousing claret shrine https://youtu.be/2WTFuSeYJ6o?si=sucecMqeSbI9qQaQ&t=1600 (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49578185) |
|
Date: January 10th, 2026 2:24 AM Author: Histrionic Hospital I think 5.2 is significantly less useful than any prior model. the safety / compliance limits are ultra sensitive and neuter it even with materially benign / open-source information. (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49578231) |
|
Date: January 10th, 2026 2:28 AM Author: arousing claret shrine It's 180 compared to Gemini 3 Pro (and loved Gemini 2.5 Pro) (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49578237) |
|
Date: January 15th, 2026 12:01 PM Author: Histrionic Hospital how's that? I started using Gemini because of how gay the safety limits are on 5.2 (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49591524) |
|
Date: January 10th, 2026 9:39 PM Author: Thirsty People Who Are Hurt Headpube its more aligned with published knowledge than ever. which is good for abstract and highly technical subject matter. horrible for anything highly social, political, uncomfortable or where human bias seeps in (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49579884) |
|
Date: January 14th, 2026 7:46 PM Author: Curious bat-shit-crazy striped hyena church building . (http://www.autoadmit.com/thread.php?thread_id=5818216&forum_id=2\u0026mark_id=5310481#49590139) |