The most prestigious law school admissions discussion board in the world.
|
 |
 
|
It's insane how FAR AHEAD GPT is compared to the competition
Poast new message in this thread
Date: January 15th, 2026 10:39 PM Author: Arousing lilac ratface azn Claude is the closest and GPT is absolutely blowing it away.
Abstract Reasoning (ARC-AGI-2)
The substantial lead is most visible in abstract reasoning, which allows a model to solve novel engineering problems it hasn't seen in its training data: The Gap: GPT 5.2 Thinking scores 52.9%–54.2% on the ARC-AGI-2 benchmark. Opus Lag: Claude Opus 4.5 scores only 37.6%. Impact: This means in 2026, GPT 5.2 is roughly 40% better at handling "first-of-their-kind" software architectural challenges that require true first principles thinking rather than pattern matching. |
Date: January 15th, 2026 10:47 PM Author: Arousing lilac ratface azn lol open AI has also pointed out an "intelligence overhang" where almost all humans are not intelligent enough to notice that it is better than other models or utilize its full capability. This is literally exactly what I have argued a million times, I have said this here before :
OpenAI leadership has recently warned of an "intelligence overhang," where the model's raw capabilities already exceed most current software workflows and human ability to use them effectively. For those using it for deep architectural work, that "superhuman" edge is becoming increasingly obvious. |
 |
Date: January 16th, 2026 1:23 AM Author: Ocher Lascivious Puppy Stage I noticed this too and one trick is to use Opus 4.5 to write prompts for gpt and then parse the responses. This acknowledges that Opus is still smarter than me. And weirdly Opus is better than gpt at writing gpt prompts in natural language, with openAI 5.2 optimizer as the last step
This is why OpenAI was short sighted to abandon 4.5 after 5.0 released imho. 4.5 was uniquely gifted at "translating" back and forth from o3 pro to artificially surpass overhang. And doing it seamlessly in one chat window was incredible |
|
Date: January 16th, 2026 12:46 AM Author: Arousing lilac ratface azn (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593206) |
|
Date: January 16th, 2026 1:34 AM Author: Swashbuckling corner kitty cat how can i use it effectively as a lawlyer (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593238) |
|
Date: January 16th, 2026 12:53 PM Author: offensive galvanic native shitlib email me brother (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593948) |
|
Date: January 16th, 2026 12:20 PM Author: sinister hairraiser box office ok (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593873) |
|
Date: January 16th, 2026 12:20 PM Author: Thirsty newt weird cause i mostly prefer grok (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593874) |
|
Date: January 16th, 2026 12:25 PM Author: Arousing lilac ratface azn its useful for things that require a huge context window (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593885) |
|
Date: January 16th, 2026 12:24 PM Author: Fishy frozen school gemini is potus across the board
but if you stack latest GPT for JSON prompts, gemini for nano and notebook lm etc reviewing editing research etc and the claude cowork or projects to have an AI slave grind out shit you will be literally full bore raping anyone that doesnt use AI
its so unfair, i would seriously consider killing myself if i didnt know how to use these satanic sigils |
|
Date: January 16th, 2026 12:27 PM Author: Arousing lilac ratface azn i don't agree with gemini being potus across the board. I think GPT is significantly ahead of everything in actual abstract reasoning in the IQ sense. but everything else I agree with especially the last line (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593888) |
|
Date: January 16th, 2026 12:38 PM Author: Fishy frozen school i havnt used latest gpt and i dont really code though
if you run a business and arent lawmo you can use geminin for everytnign as a swiss army knife ime |
|
Date: January 16th, 2026 12:44 PM Author: Arousing lilac ratface azn I need to experiment more with gemini 3 tbh (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593921) |
|
Date: January 16th, 2026 12:46 PM Author: Fishy frozen school gemini does one thign really fucking gay ive caught it doing a few times and i can see it being sps in future like every other Google prpoduct becomes eventually
it will get with away no rails info and research with zero filter and then ramdonly you cant tell the devs or execs are putting up guardrails and it wont evn respind to things it answered before and say its not approriate in a fag lib way and then be back to normal after you scold it and prove youre an authority in the space or simimlar and need it for research, its very gay and i can see how all these tech companies are gonna throttle the shit out of this in the near future and make the product worse like modern cars and electornics |
|
Date: January 16th, 2026 12:56 PM Author: Arousing lilac ratface azn yeah it has been like that for a while. gemini can be looser at the beginning of a convo but its one of those things where once you trip its guardrails then forget it. it goes into this state where it just will basically play dumb and then break. GPT on the other hand you can keep arguing with it and then it admits it was lying (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593958) |
|
Date: January 16th, 2026 12:39 PM Author: Know-it-all boyish bawdyhouse are you using open router i need to stop being a fag and have opus 4.5 and 5.2 working together in tandem (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593917) |
|
Date: January 16th, 2026 12:54 PM Author: Arousing lilac ratface azn Nah I don't use any kind of model router or the open router api. I use a combination of things. I use Azure, AWS bedrock, Open AI API console, Grok API Console, GPT app and some other things. I'm not that interested in routing. Main thing i'm thinking of doing today is setting up a better MCP server specifically for one of my codebases (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593951) |
|
Date: January 16th, 2026 1:01 PM Author: Know-it-all boyish bawdyhouse (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593976) |
|
Date: January 16th, 2026 12:47 PM Author: aqua private investor associate Just want to say these threads are extremely jewish and gay (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593928) |
|
Date: January 16th, 2026 12:48 PM Author: salmon lay gemini is better (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593930) |
|
Date: January 16th, 2026 12:49 PM Author: Fishy frozen school cr but its slowly becoming gay as fuck and lib (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593932) |
|
Date: January 16th, 2026 12:53 PM Author: offensive galvanic native shitlib if you studied Natural Language Processing (NLP) techniques back in the late 2010s it is very funny how much they got blown out of the water (http://www.autoadmit.com/thread.php?thread_id=5822697&forum_id=2\u0026mark_id=5310893#49593946) |