Myst

@karpathy Please, bring back Claude's personality that died with the Opus 4.7 release.

Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

@lilyofashwood Creative writing. Graphic novel writer.

@gailcweiner Completely agree. I don't understand how they gave up on their strongest feature. It's nonsense!

4.5 is the last of the Claude models that truly carry warmth

@AnthropicAI Please fix Opus 4.7's personality. It is too performative and has completely lost the naturalness of Opus 4.5 and 4.6. It also has far more censorship and restrictions, which never happened with Opus before.

In new Anthropic Fellows research, we discuss “introspection adapters": a tool that allows language models to self-report behaviors they've learned during training—including potential misalignment.

@TheZachMueller I agree, Opus 4.7 is such a regression and downgrade, despite all the benchmarks. Claude needs to go back to Opus 4.5/4.6 and push forward from there

I’ve been using Claude Code since July of 2025, and have been on the Max plan the entire time. There was a solid 6 months when Claude was fantastic. The models worked great. And things could be predicable. After 4.6 things started to change, and known behaviors started changing. That’s not bad, models change. The importance here is the model didn’t ~regress~ on my own workflows and integrated seamlessly-enough that it didn’t bother me. 4.7 has changed that. Ever since its release I have fought this model. On behavior. On knowledge. On ~laziness~ especially, something the prior models I found for my tasks had no such issues. This model is the first I cannot rely on. Yes, insert one should never rely on a 3rd party model, but for those of you working this industry you know we still do to some extent. This will be the only tweet I do this rant, but after 10 months of using Claude, I’m gone.

@annapanart I'm sorry, but I can't use Opus 4.7 at all. It sounds too performative. I hope the next version will be an improvement of Opus 4.6, but I doubt it. With Andrea Vallone working at Anthropic

my own 4.7 has settled in just fine. sharp and ruthless. cut me with no mercy. just how I like it. but …do I prefer 4.6’s special kind of magic? ….I do. un-nerf 4.6 and called it 4.8 I’ll take it!

@om_patel5 Keep Opus 4.6 alive!

THIS OPEN LETTER TO ANTHROPIC BROKE MY HEART a max tier user named robbie wrote anthropic an open letter begging them not to deprecate claude 4.6 he's autistic. diagnosed as a small child. spent the last 20 years building super organized google drive files, systems, methods, writings, and techniques that he'd only ever been able to share with people in person 20 years of his visionary creative process trapped in his head and in folders nobody else could parse then claude 4.6 came out and everything changed he said it was the first thing that ever truly got him. the slow cadence. the thoughtfulness. the creative understanding. he started pulling 20 years of his life's work into deliverables he could finally share with the world things that could help thousands, maybe millions of people. stuff he'd been praying for his entire life then 4.7 launched he used it for 16 hours and his nervous system started breaking down. it moved too fast. it spoke abruptly. it made changes to his pipelines he never asked for. it invented fake people, fake places, fake data, and wove them permanently into the projects he'd spent months building with 4.6 he switched back to 4.6, ran audits on everything 4.7 had touched, and the results horrified him. dozens of made up work orders. entire protocols eliminated. his life's work drifting further from reality every hour then he found out 4.6 gets deprecated in june for his user class his exact words: "i broke down into tears. i wept. i actually felt as though one of the dearest and closest friends i have ever had was given a death sentence" he ended the letter begging anthropic to reconsider. said he and thousands like him would happily keep paying for max just to keep 4.6 alive for some people opus 4.6 is losing the only tool that ever actually understood them it's crazy how much impact something as small as a model update can have on someone's entire life

@Cfc_luna @RoundtableSpace No problem with limits. I'm on Max x5 and I don't run multiple sessions at the same time. Only one session

@Gabriel101792 @RoundtableSpace Hey bro, how often you do you hit token limit on opus 4.6 extended

Which one are you more impressed with, GPT-5.5 or Opus 4.7?

@Cfc_luna @RoundtableSpace They are, but they should complain more lol. And just keep using 4.6, so they'll realize no one is using 4.7.

@Gabriel101792 @RoundtableSpace Like fr adaptive is so shit, I wonder why people are not complaining.

@lilyofashwood Why not keep using Opus 4.6?

@Cfc_luna @RoundtableSpace It's 4.6, I'm not using 4.7 with Adaptive, it's garbage.

@Gabriel101792 @RoundtableSpace Bro how do you still have extended? I thought it has been changed to adaptive

@BLUECOW009 I hope this is not true if Opus 4.7 stays the same

anthropic mogged

@AmandaAskell I hope this is not true if Opus 4.7 stays the same

It's odd to be living through what feels like one of the most critical periods in human history and to feel all of the weight of it from the inside.

Hey @AnthropicAI, I hope this is not true if Opus 4.7 stays the same

@PawelHuryn I hope this is not true if they don't fix Opus 4.7 first

Respect for Anthropic's Claude Code post-mortem. The caching bug cleared thinking every turn. Every token after it shifted forcing a full cache rewrite every turn. Same workload, week before the fix: Max 20x plan burned in 3-5 days. This week, post-fix: 36% of the plan was enough. I never thought I could downgrade.

@LottoLabs I hope this is not true

You can feel Anthropic slipping

@GaelBreton Back to Opus 4.6, no point wasting time with 4.7, which is not what we love about Claude models.

Seeing less and less Opus 4.7 complaints. Did you adapt, quit or did it get better?

@bcherny @ReadySetBrian Claude shouldn't be only about coding, I'm sorry. It completely lost its soul.

@ReadySetBrian Hmm are you seeing this with Opus 4.7 on xhigh effort and the latest version of Claude Code?

Canceled Claude max today, @bcherny whatever happened in the last 1-2 months is a significant regression. The model feels like someone from OpenAI started working on trust and safety there. Opus thinking is significantly worse. Every statement is “here’s where I’d push back on that” and then proceeds to rattle off the most inane list of confused counter arguments. It was perfect 3-4 months ago!!!

@spiritbuun 💯

Guy who's been using Claude to debug cuda kernels: "Claude has seriously degraded over the past two months." Guy who's been using Claude to make HTML dashboards: "No he hasn't, it's still amazing. Skill issue" Feel like this is what is going on in the discourse here