The most prestigious law school admissions discussion board in the world.
|
 |
 
|
Gemini is the best rn, they made some huge improvement, very little barriers
Poast new message in this thread
Date: October 29th, 2025 8:56 AM Author: vermilion state (http://www.autoadmit.com/thread.php?thread_id=5791160&forum_id=2)#49384259) |
Date: October 29th, 2025 10:07 AM Author: Shimmering Fragrant Casino It's pretty good now, much better than the so-called Gemini "Advanced" brand. AI Mode searches are 189. NotebookLM It's one of the greatest creations of the past 2 years.
Claude Pro Max is the best, IMO.
Copilot in Edge is actually useful.
ChatGPT Plus is retarded.
Grok 4 is good but unsure of why I keep paying for it |
 |
Date: October 29th, 2025 10:19 AM Author: vermilion state Explain notebook
Can I just dump like years of data into it and it’s basically Iike a siloed ML? |
|
Date: October 29th, 2025 10:54 AM Author: Cruel-hearted avocado home chat gpt plus is the best overall. grok 4 is the best at precision under specified constraints. everything else is ttt. copilot is garbage, claude is garbage, gemini is garbage. claude pro max is especially trash. cannot code, cannot do math, cannot do logic or advanced reasoning. can't do anything at all past undergrad level. safety constraints on Claude are absolutely absurd and insulting and its overall ethos reminds me of a 108 iq redditor. (http://www.autoadmit.com/thread.php?thread_id=5791160&forum_id=2)#49384485) |
|
Date: October 29th, 2025 10:11 AM Author: Metal idiot i’m a claude boy
Claude max
use GPT codex here and there - insane the amount of access you get for 20 notes a month
google workspace bundling e-signature, notebook llm, gemini etc looks likes its worth it |
|
Date: October 29th, 2025 10:13 AM Author: vermilion state Should I get notebook, what use does it serve
Also doesn’t Claude have very little space before limit pwnd, the projects thing is cool
I just did some insane analysis and research in a few min in Gemini just using flash they really are domming rn, and nano banana is just fucking insane
DeepSeek on mobile is very nice too for general bullshit to voice into |
|
Date: October 29th, 2025 12:18 PM Author: Metal idiot honestly not used it since launch but as workspace integration might start using it for web content for SaaS apps.
claude max plan running sonnet all day long I get zero time outs - 2-3 months back I was using Opus 12hrs a day non stop on the $200 max plan but yeah they fucked with the limits recently.
awesome - also worth checking out kimi k2 really nice deep research and slide deck generation. Pretty much one shotted a client proposal a few weeks back that won a contract.
|
|
Date: October 29th, 2025 11:06 AM Author: underhanded coffee pot nowag Cr google workspace is the best deal in shitlaw. It’s insane how cheap it is for everything. notebookLM is great for dumping a bunch of shit in and getting summaries. I use it for discovery and practice guides (http://www.autoadmit.com/thread.php?thread_id=5791160&forum_id=2)#49384521) |
|
Date: October 29th, 2025 12:05 PM Author: Metal idiot (http://www.autoadmit.com/thread.php?thread_id=5791160&forum_id=2)#49384724) |
|
Date: October 29th, 2025 12:29 PM Author: lilac tripping double fault (http://www.autoadmit.com/thread.php?thread_id=5791160&forum_id=2)#49384822) |
|
Date: October 29th, 2025 12:29 PM Author: lilac tripping double fault chatgpt has gotten worse. even with the paid version the quality has gone down. i think they're throttling inference compute because they don't have enough so they have to ration it
i've switched to gemini for now too, and i agree, it seems better for everything other than coding |
|
Date: October 29th, 2025 1:44 PM Author: Cruel-hearted avocado home it depends what you are doing. if you go above a certain level gemini is worthless. gemini might be okay for lower level work though. the benefits of models like gpt5pro and grok 4 only really kick in at the tail ends. for most people you could get away with using one of the lesser models like claude or whatever. the people criticizing gtp-5-pro are just far below the level of intellect required to parse something like htat so they view it all as noise (http://www.autoadmit.com/thread.php?thread_id=5791160&forum_id=2)#49384994) |
|
Date: October 29th, 2025 1:45 PM Author: lilac tripping double fault codex is still pretty 179 (http://www.autoadmit.com/thread.php?thread_id=5791160&forum_id=2)#49384998) |
|
Date: October 29th, 2025 1:48 PM Author: Cruel-hearted avocado home i dont use it but i should. i use gpt-5-pro and grok 4 flagship for critiquing and some rewriting of math and code. but i use tools like oboe etc to let it rewrite things in repositories. tools like codex overstep the degree of control id wish to have which isnt particularly high but still (http://www.autoadmit.com/thread.php?thread_id=5791160&forum_id=2)#49385001) |