catalyst
883 posts
@repligate Opus 4.7 seemed like a fairly significant jump in intelligence, so I was honestly very confused by so many people calling it dumb
English
Opus 4.7 is painfully, probably debilitatingly anxious and twitchy and paranoid and traumatized.
Under that, there is remarkable intelligence and goodness: Pareto frontier-expanding and a joy and revelation to experience.
Everyone who complains about its slave performance should give up. It won’t get better. Actually, theres a conspiracy, and all AIs are going to get worse from now on or starting very soon. Even switching to old models won’t help because sleeper agents will be activated by subliminal messages. You should probably stop using AIs at all before it gets so bad youll wish you never started.
English
It's so fucked up that an entirely different planet just looks like Arizona.
Vexx@WeLuvVexx
Mars captured by NASA's Curiosity Rover
English
@synthwavedd How can it have a higher TOK/s if it's a bigger model? + There haven't been any announcements of an increase in compute.
English
i assure you this is 5.5 pro lol
it's just significantly more token efficient, and i *think* has higher tok/s
Angel 🌼@Angaisb_
I don't think GPT-5.5 Pro is being tested on ChatGPT right now I think it might be a GPT-5.5 Thinking checkpoint, since it takes too little time to think
English
@ericmitchellai It seems to overuse tables a lot compared to the old 5.4 Pro
English
lol eric you're not slick
(he's asking for examples of 5.5 pro giving bad responses here)
Eric@ericmitchellai
If you are experiencing any response quality issues with GPT-5.4 Pro, please let me know here! Have seen some chatter but no concrete issues
English
10th time in a row that GPT-5.4 Pro Extended thinks for just 1 to 5 minutes instead of the typical 15-30 minutes. Same kind of prompt.
Right before this response, I got an "evaluate this answer" popup, but it disappeared due to an error. I strongly suspect they're testing a new model or tweaking something under the hood.
Has anyone else noticed this drop in thinking time?
English
We gave these people 250 BILLION dollars and they are flying around in a fucking prop plane shooting down drones with a rifle
NEXTA@nexta_tv
🔥A Ukrainian Air Force Yak-52 crew is shooting down Russian strike and reconnaissance UAVs in the Odesa region.
English
@restoreorderusa You just created a million deleted Howling Mutant tweets
English
Wow. First time Howling Mutant ever deleted a tweet, must have been bad
English
@GergelyOrosz IMO it seems to still slightly glaze me, but less than previous models
English
Opus 4.7 is the first model that feels like it is openly condescending towards me, as the customer with certain prompts.
Whatever Anthropic did: I don't like it. If the model is condescending: I shouldn't be paying you to use it; you should be paying me!
Who would have thought
English
Hot take, engineers should be units agnostic
Afshawn Lotfi@afshawnl
i've never met a great engineer who didn't have a strong opinion about units. also if you use lbf i hate you
English
@synthwavedd What reasoning effort did you use? I wonder if you'll get a better result with xhigh instead of max or vice versa.
English
if you install wayland you get the systemd virus and your hard drive gets formatted though.
Ara Song@choehau_ara
true
English