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

Opus 4.5 is disgustingly good

Pretty sure humans are effectively useless now other than to...
saffron magical rigpig
  11/29/25
The level of improvement from Sonnet 4.5 is insane (a model ...
passionate translucent tanning salon
  11/29/25
yeah and whats also interesting is that sonnet 4.5 was legit...
saffron magical rigpig
  11/29/25
...
judgmental kink-friendly theatre liquid oxygen
  11/29/25
My retarded fucking company refuses to upgrade our GHCP mode...
Filthy Heaven
  11/29/25
lol jfc
saffron magical rigpig
  11/29/25
...
diverse geriatric stage
  11/29/25
wow
crimson alcoholic step-uncle's house
  12/07/25
lmao
Wonderful Doobsian Becky
  12/07/25
...
diverse geriatric stage
  11/30/25
it truly is insane - what a gift
crimson alcoholic step-uncle's house
  12/07/25
explain what cause you to reach this conclusion
know-it-all water buffalo
  12/07/25
Mental illness
At-the-ready French Chef
  12/07/25
I need an AI agent to work with
ultramarine gaping mediation
  12/07/25
Opus 4.5 is up to 4 hours and 49 minutes on the METR time ho...
passionate translucent tanning salon
  12/21/25
People complaining that 4.5 was shadow throttled starting a ...
Demanding locale partner
  12/21/25
They are always claiming this for new model releases and no ...
judgmental kink-friendly theatre liquid oxygen
  12/21/25


Poast new message in this thread



Reply Favorite

Date: November 29th, 2025 12:55 PM
Author: saffron magical rigpig

Pretty sure humans are effectively useless now other than to pilot AI systems

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49470203)



Reply Favorite

Date: November 29th, 2025 1:37 PM
Author: passionate translucent tanning salon

The level of improvement from Sonnet 4.5 is insane (a model that is a couple months old). LLM progress isn’t really slowing down in a meaningful way.

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49470302)



Reply Favorite

Date: November 29th, 2025 2:10 PM
Author: saffron magical rigpig

yeah and whats also interesting is that sonnet 4.5 was legit good..like it was better than opus 4.1 for most things and competitive with gpt5. I still like gpt 5.1 and the 5 pro api best as a generalist but opus 4.5 is insane for coding and also its much better at critiquing something you give it. the other LLMs are still too nice, but it has no problem giving harsh criticism and arguing against you. GPT 5.1 is better than 5 for this but still a bit weak compared to claude

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49470392)



Reply Favorite

Date: November 29th, 2025 2:12 PM
Author: judgmental kink-friendly theatre liquid oxygen



(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49470405)



Reply Favorite

Date: November 29th, 2025 2:13 PM
Author: Filthy Heaven

My retarded fucking company refuses to upgrade our GHCP models from gpt4o and I want to kill myself

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49470410)



Reply Favorite

Date: November 29th, 2025 2:49 PM
Author: saffron magical rigpig

lol jfc

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49470540)



Reply Favorite

Date: November 29th, 2025 2:50 PM
Author: diverse geriatric stage



(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49470543)



Reply Favorite

Date: December 7th, 2025 10:17 AM
Author: crimson alcoholic step-uncle's house

wow

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49491040)



Reply Favorite

Date: December 7th, 2025 10:32 AM
Author: Wonderful Doobsian Becky

lmao

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49491074)



Reply Favorite

Date: November 30th, 2025 1:14 AM
Author: diverse geriatric stage



(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49471468)



Reply Favorite

Date: December 7th, 2025 10:19 AM
Author: crimson alcoholic step-uncle's house

it truly is insane - what a gift

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49491048)



Reply Favorite

Date: December 7th, 2025 10:23 AM
Author: know-it-all water buffalo

explain what cause you to reach this conclusion

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49491063)



Reply Favorite

Date: December 7th, 2025 10:50 AM
Author: At-the-ready French Chef

Mental illness

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49491106)



Reply Favorite

Date: December 7th, 2025 10:44 AM
Author: ultramarine gaping mediation

I need an AI agent to work with

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49491097)



Reply Favorite

Date: December 21st, 2025 12:59 PM
Author: passionate translucent tanning salon

Opus 4.5 is up to 4 hours and 49 minutes on the METR time horizon task. this benchmark measures the task length (in terms of human work time) that models can do with SWE/AI research type projects. big increase over 5.1 max, which was 2 hours 53 minutes and faster than the overall trend of doubling every 7 months. with any luck, models will be capable of substantially automating AI research before 2030 and set off an intelligence explosion.

https://metr.org/blog/2025-03-19-measuring-ai-ability-to-complete-long-tasks/

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49527643)



Reply Favorite

Date: December 21st, 2025 1:47 PM
Author: Demanding locale partner

People complaining that 4.5 was shadow throttled starting a few days ago

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49527733)



Reply Favorite

Date: December 21st, 2025 1:53 PM
Author: judgmental kink-friendly theatre liquid oxygen

They are always claiming this for new model releases and no objective evidence ever materializes for it. It’s almost certainly a psychological bias rather than reality. There is substantial variation in how models respond to a particular problem just based on how they are prompted. It’s hard for a user to reliably measure model capabilities over time based on intuition alone.

(http://www.autoadmit.com/thread.php?thread_id=5804093&forum_id=2\u0026mark_id=5310893#49527743)