The most prestigious law school admissions discussion board in the world.
|
 |
 
|
Gemini 3 is going to be the POTUS of LLMs
Poast new message in this thread
Date: November 13th, 2025 10:38 PM Author: .,.,,..,..,.,.,:,,:,...,:::,...,:,.,.:..:. https://youtu.be/oeWPC_bC3zs
crushes Sonnet 4.5 at coding. not even remotely close. almost certainly going to dominate every other LLM on the market. be prepared to ditch OpenAI and Claude |
Date: November 13th, 2025 10:40 PM Author: mr. lowercase i already started using gemini over chatgpt 5 anyway
the compute throttling by openAI is so ridiculously jewish/indian, i'm not paying money for that shit |
 |
Date: November 13th, 2025 10:42 PM Author: Mainlining the $ecret Truth of the Univer$e (One Year Performance 1978-1979 (Cage Piece) (Awfully coy u are)) (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49429061) |
|
Date: November 13th, 2025 10:46 PM Author: .,.,,..,..,.,.,:,,:,...,:::,...,:,.,.:..:. Gemini 2.5 pro is slightly dumber than GPT-5 but seems to write much better. GPT-5 is not optimized for clarity. (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49429067) |
|
Date: November 13th, 2025 10:57 PM Author: https://imgur.com/a/o2g8xYK lol (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49429085) |
|
Date: November 13th, 2025 11:05 PM Author: ,.,...,.,.,...,.,,. So much for LLM scaling hitting a wall. (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49429096) |
|
Date: November 14th, 2025 5:36 PM Author: .,.,,..,..,.,.,:,,:,...,:::,...,:,.,.:..:. https://generativehistory.substack.com/p/has-google-quietly-solved-two-of
yeah, this looks much more impressive than the GPT 5.1 garbage |
|
Date: November 14th, 2025 5:39 PM Author: Mainlining the $ecret Truth of the Univer$e (One Year Performance 1978-1979 (Cage Piece) (Awfully coy u are)) wow (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49431389) |
|
Date: November 14th, 2025 5:39 PM Author: Mainlining the $ecret Truth of the Univer$e (One Year Performance 1978-1979 (Cage Piece) (Awfully coy u are)) https://research.google/blog/introducing-nested-learning-a-new-ml-paradigm-for-continual-learning/ (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49431392) |
|
Date: November 14th, 2025 5:44 PM Author: the forced removal of evil from power Gemini started doing some fucking gay against its terms/policy shit on me lately especially with nanobanan drawing of niggers kikes and chinks (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49431415) |
|
Date: November 18th, 2025 11:11 AM Author: .,.,,..,..,.,.,:,,:,...,:::,...,:,.,.:..:. https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-Pro-Model-Card.pdf
LJL at OpenAI and Anthropic and their tiny pink models. |
|
Date: November 18th, 2025 11:11 AM Author: UhOh can't believe i didn't buy goog at 150. it was way under its peers for no reason. fml. (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49440506) |
|
Date: November 18th, 2025 11:17 AM Author: .,.,,..,..,.,.,:,,:,...,:::,...,:,.,.:..:. (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49440518) |
|
Date: November 18th, 2025 11:26 AM Author: .....;;,,.........;.;.;.;.,;,;,;.;.;,; The theory has always been that they had an insurmountable data advantage from 2+ decades of hoovering up all their segment dominating app data, but the longer that took to show up the harder it was to stick with the thesis. I will say that their product development/execution has fallen off a cliff and they went so hard with the H1B indentured servitude that it’s permanently damaged their reputation with talent, and that could be the MSFT ace in the hole. The median highly sought after super brain early 20s recent grad tech dork (who is a U.S. citizen) looks down his nose at a Google job v 15 years ago when it was the best outcome possible. Also Google cloud is a joke v MSFT’s being legit and AWS being dominant. The main funnel for Claude is just the AWS Bedrock channel. I suspect we end up with three main players just like cloud, but Google may end up with a significant long term advantage v the others if they don’t have to worry about a separate company that can fuck up and screw their strategy. I do think the the next 2-3 years are the moment of clear advantage for the wait and see/“fast” follow strategy, and in that sense AWS still has yet to really make its move, but they’re maintaining a foothold just with one situational partnership. (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49440552) |
|
Date: November 18th, 2025 12:17 PM Author: ,.,...,.,.,...,.,,. they likely also have more compute available than other players. they have been working on their own chips for a while and prioritized AI sooner than other players. they also have a ton of talent at DeepMind. they weren't prioritizing LLM products for a while but now that they are, they'll likely to continue to churn out products on the cutting edge. these benchmark scores are pretty far ahead of their competitors. (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49440664) |
|
Date: November 18th, 2025 12:38 PM Author: Mainlining the $ecret Truth of the Univer$e (One Year Performance 1978-1979 (Cage Piece) (Awfully coy u are)) (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49440761) |
|
Date: November 18th, 2025 12:47 PM Author: .....;;,,.........;.;.;.;.,;,;,;.;.;,; I am not sure the DeepMind talent really compares and has the same degree of capability in terms of building what will sell as opposed to the classic Google idealism of wanting to build the coolest thing they can come up with. The thing about AWS and the talent raids that MSFT and OpenAI have been pulling off for the last two years is that they have hoarded so much of the talent that has access and established relationships with key contacts who make the buying decisions throughout the F500 plus long track records of identifying and attaching to the hyper growth startups. Google has always sucked at that and sat a distant third in cloud for 10+ years as a result. The freemium non-enterprise path is also massively complicated by the giant lead in brand awareness and user adoption that ChatGPT has already built out. It’s possible that Gemini 3 will be so vastly superior that it will be like Gmail dropping and killing off Hotmail etc., but it’s also very possible that they end up building a Betamax ie a superior technology that gets left behind because it can’t find a path to revenue and profit. (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49440790) |
|
Date: November 18th, 2025 12:55 PM Author: Mainlining the $ecret Truth of the Univer$e (One Year Performance 1978-1979 (Cage Piece) (Awfully coy u are)) Recent daily and monthly stats have been very favorable for Gemini, relatively.
I'm sure you've seen the charts.. U see Google slowly creeping up and up..plus Anthropic (via its enterprise accounts). |
|
Date: November 18th, 2025 3:24 PM Author: ,.,...,.,.,...,.,,. The presence on Google search ensures many will use it even if Google doesn't market well outside of it. Word will get around that it is better than chatgpt (http://www.autoadmit.com/thread.php?thread_id=5797692&forum_id=2#49441296) |