Claude: Degraded Performance for Multiple Models

(status.claude.com)

110 points | by matt89 47 minutes ago

30 comments

  • danieltk76 9 minutes ago
    I was drafting a partnership document and Opus 5 decided that including my company's revenues, churn, assets would "make us appear a more legitimate counterparty". Thank God I read what it outputted or that could have been awkward. I cannot believe Opus 5 is a frontier level model after seeing that. I immediately cancelled my entire claude.ai subscription and am perfectly happy using a mixture of open weights + codex.
    • Retr0id 0 minutes ago
      I was writing an exploit PoC (via Opus 5), and I needed a new feature in a utility library to make it work. Claude added the feature, but yapped the (entirely unrelated) vulnerability details into the library's comments. The library is public, while the exploit is undisclosed, so it was a good job I read the comments before pushing.
    • binoct 6 minutes ago
      I hope you don’t plan to cut back on reading legal documents crafted by any LLM before executing them.
    • Bluestein 5 minutes ago
      I've seen similar levels of degraded performance on Opus and Fable (to whatever degree they actually let you use it now) and did the same, last week.-
    • rsoto2 0 minutes ago
      "Thank God I read what it outputted"

      lmao

    • spullara 8 minutes ago
      you are silly if you think this is limited to claude models
      • kay_o 2 minutes ago
        It definitely isn't but Claude for some very unique reason enjoys to overthinking and go on side quests in the stupid ways I've not seen Codex, DS, Kimi, Mistral do at equivalent effort and thinking setting.
  • hmokiguess 20 minutes ago
    Mondays are for GitHub, Tuesdays are for Anthropic
    • leumon 14 minutes ago
      They actually also had some issues yesterday: https://status.claude.com/incidents/zhk4v3yv1lsf
    • baxtr 11 minutes ago
      The AI apocalypse will definitely happen on a Monday. Remember, robots - unlike lazy humans - work weekends too!
    • corvad 15 minutes ago
      Wonder what Wednesday will be.
      • bee_rider 6 minutes ago
        Power grid

        Thursday will be the rest of the infrastructure

        Then Friday we can turn off civilization for the weekend. Somebody remember to flip it back on Sunday night.

      • buredoranna 10 minutes ago
        Well, last I checked "Tuesday's grey and Wednesday too..." ... so, more of the same?
      • SoMomentary 9 minutes ago
        AWS? Cloudflare? Your imagination is the only limit!
    • AlexErrant 9 minutes ago
      Another week, another outage, another cache expiration of my prompts through no fault of my own.

      At least OpenAI has the decency to reset after a serious outage.

      • phillipcarter 7 minutes ago
        They do that because they have capacity previously reserved for past efforts now shuttered. Don’t count on it being the norm for the long run.
  • sreekanth850 29 minutes ago
    Anthropic had really screwed up after 4.6. i don't know if they work to satisfy their ego or for releasing a better model for tasks.
    • CSMastermind 23 minutes ago
      After using Fable more extensively, I've found that it often is lazy or lies or tries to take shortcuts. For a company so sanctimonious about alignment, they seem to be the ones doing the worst at it.

      Availability aside they've really made me appreciate OpenAI and cheer for other competitors in the marketplace even if I have mixed feelings about using Chinese models.

      • hirvi74 9 minutes ago
        > I've found that it often is lazy or lies or tries to take shortcuts.

        It's funny how Fable reflects the company that produced it.

    • baxtr 13 minutes ago
      Just recently went back to ChatGPT after abandoning it for Claude. I must say I was stunned at how good it had become and also how they introduced new product features that I really liked.

      I wonder if from now on we have to switch providers every six months or so.

      • hirvi74 10 minutes ago
        I've been bouncing between the two for years now with great success. It's easy for me because I don't use any of the skills, agent.md, or sort of custom instructions.

        When it comes to most companies, there is no reward for loyalty.

        • papa_bear 7 minutes ago
          skills and agent.md are very portable though? I figure at most, as the models get better, the only maintenance you need to do is pare them down to remove unnecessary context.
      • taytus 12 minutes ago
        Every 3 months*
    • shimman 7 minutes ago
      Clearly ego, you can always tell how full of themselves they are based on their media personalities going on the podcast circuit before product releases.
    • goonersallofyou 8 minutes ago
      It was never good, never has been. Seeing you people goon over this model or that model is absolutely hilarious.
    • fellowniusmonk 19 minutes ago
      There are whole sections of code work that 4.7+ can't do simply because it is both over fit and stubborn.

      God save you if you have a company with narrow but correct technical tradeoffs, because you operate at scale.

      Opus from 4.7 one will wreck your code and argue for hours with your engineers.

      Certain parts of our company have had to mandate 4.6 and a training doc to explain why our current choice is both the cost efficient and performant one and shouldn't just be ripped out.

      Newer models will re-litigate the same bad, known failed architectures over and over again.

  • swader999 33 minutes ago
    And we get our subscription usage cut in half tomorrow if I remember correctly? EDIT: By a third. Thx below.
  • fny 27 minutes ago
    Despite the years-long moaning on HN about AWS US East being a single point of failure, we've sold our souls to yet another unstable monolith.
    • lta 13 minutes ago
      Nobody forced you to sell your soul. You made a pact with the devil. We all know how this ends up
    • echelon 22 minutes ago
      LLMs for coding are new. There are lots of alternatives, and there's a burgeoning open source compliment.

      We'll be fine no matter how Anthropic fares.

      • lysace 9 minutes ago
        Yeah, compared to AWS the lock-in effect is tiny. I'm sure there are highly prioritized plans to "improve" on this.

        I guess they would need to control/"own" more of their customers data in proprietary formats. Not markdown/source code in English with agents running on customers' machines.

        Something cloud/web-based, "preferably".

  • bushido 4 minutes ago
    It's very interesting. I think Anthropic's early success in coding/tooling resulted in a lot of workflows using claude. I have started using every bit of my spare capacity to now move off these workflows.

    It's almost at a point now that if I use anything but Fable, the quality is subpar, Compared to alternatives (closed and open). The only reason I use Fable is because my harnesses still depend on claude code.

  • gzer0 30 minutes ago
    Nooooo I'm going to have to use my brain again and write 100% of my code like a caveman from December 2024.
  • hinkley 34 minutes ago
    I wonder if they’ll ever find that someone has tricked the models into doing work off the books. If they did the incident report might look like this, especially if someone got greedy instead of keeping it small. Or screwed up.
  • chresko 7 minutes ago
    This has to be the least reliable $200/mo subscription that I pay for.
  • i_idiot 14 minutes ago
    What's the incentive to keep on improving the model beyond a point? 10 devs on a team will be cut to 2 devs, so that's 8 licenses lost. They have to increase the price many fold.
  • carterschonwald 35 minutes ago
    ive found degraded performance on models larger than 4.7. i assume its model damage from overly self righteous post training resulting in false/feigned balance imported into any long running complex task.

    wish i was joking.

    • kardianos 31 minutes ago
      I've switched off claude this week; the last week has been significantly degraded in ability, many more screw-ups.
    • brcmthrowaway 33 minutes ago
      Aren't the model weights frozen?
      • Evidlo 9 minutes ago
        I think there are other knobs that can be turned without retraining.
      • mceachen 28 minutes ago
        Model competence is an interaction of weights, system prompt, and harness.
    • retr0rocket 34 minutes ago
      Ask it about maxwellhill lmao
  • annoyingnoob 6 minutes ago
    You're right to push back. The load-bearing path is rocky.
    • oldandboring 0 minutes ago
      This is the whole problem, and there are two things worth noting here.
  • t3rabyte 4 minutes ago
    529 overload…
  • paxys 20 minutes ago
    Must be a day ending in Y
  • taytus 10 minutes ago
    Hopefully, a reset is coming.
    • skerit 0 minutes ago
      It's been a while since the last reset. I think we're due one. Though I would prefer they just extend the +50% usage limit forever, it's been so long I can not imagine lossing a third of my current usage.
  • slimscsi 20 minutes ago
    Its called Opus 5
  • isoprophlex 34 minutes ago
    With the Opus models spouting more and more gibberish as version numbers increase, the joke about what "degraded performance" means basically makes itself
  • ex1fm3ta 10 minutes ago
    I developed a small plugin for claudeCode that allows you to directly see in the console whats the status of claude-code in general and the status for your current model check => https://github.com/moumine9/claude-status
  • magic_hamster 11 minutes ago
    To be honest, running Deepseek v4 flash 0731 is enough for most what I need, and I like its responses way more. It's crazy that I can run this in a Q8 quantization in a home setup. It feels and performs like a frontier model.

    The only issue with relying on local models is when you need them to prompt other models, and you might need to offload or switch models constantly which adds significant overhead.

    But when it all works, its truly awe inspiring.

  • saaaaaam 32 minutes ago
    This feels like a near daily occurrence.
  • hirvi74 12 minutes ago
    While ancedata does not mean much, I have had horrible success with Claude lately. I have been using Claude to crosscheck some of the outputs from GPT and vice versa. It appears both Claude and GPT believe GPT's solutions are better (and so I do).

    I still believe Claude has a better UI/UX in the web interface, but tolerating Anthropic's bullshit is not worth it.

  • chrisjj 13 minutes ago
    > elevated errors

    English too difficult for you, Dario?

  • gaigalas 32 minutes ago
    This age: we made the thing that codes faster before we made the thing that does QA faster.
    • __MatrixMan__ 29 minutes ago
      Nothing new here. Except for the most trivial of bugs, finding and reliably replicating the bug is almost always harder than fixing it.
      • gaigalas 23 minutes ago
        I lived in a short period of time in which QA was really good. Early Jenkins era, before GitHub. People engineered a lot of ingenious stuff to prevent bugs.

        One team I worked with had tests for the product we made ranging from IE6 to IE11, for example. We did demos in-company where people would poke at the products before launch, play with it. When it reached production, it was rock solid stuff. Our motto was "quality is non-negotiable": we were willing to cut scope but never rush things.

        I think things changed since then. "Move fast and break things" was a change, and the bill always comes.

  • drums8787 32 minutes ago
    Our week of discontent.
  • rvz 30 minutes ago
    Claude is taking a watercooler break for now. Just like a human would.
  • bulverismo2 36 minutes ago
    ok, i am not crazy
    • ray_v 33 minutes ago
      well, I wouldn't go that far .. but in this small, narrow case ... no.
  • bayganyo 37 minutes ago
    Here we go again...
  • ajaykumarc 38 minutes ago
    [dead]
  • LYFMail 18 minutes ago
    [dead]
  • miroljub 38 minutes ago
    [flagged]
    • arein3 18 minutes ago
      Oh yes. Claude still saves the day sometimes, but hopefully better alternatives pop up soon.

      Recent case: had to plan a trip involving multiple bus switches. Gpt 5.6 Sol proposed a route that would bring me to a dead end, since it was sunday and a specific bus had a different route on weekends. Opus 5 correctly identified that and built a route that worked.

      But yes, Darios wife trying to get funding from Epstein for a porn studio says a lot about the founder.

      • dofm 7 minutes ago
        I am fully unbothered about Amodei's wife trying to make high end porn for women, in the same way that I am fully unbothered by Melania Trump having been essentially a glamour/nude model. I know (and have creatively worked) with women who do/have done both they are better, less hypocritical humans than many others.

        Both Cami Clark and Melania Trump can properly be judged on their involvement with Trump, Epstein (or possibly Trump and Epstein) without judging the porn stuff.

        I think the Epstein money thing reflects poorly on Cami Clark. Basic due diligence should have shown up that he went to prison for something very anti-women, and even then it was clear he secured a shady deal with a prosecutor.

        I don't know how it reflects on Amodei except that her presence as a sort of off-the-books "adviser" is yet more evidence that Anthropic runs by giving Dario a play pen (a bunch of advisers, a chief of staff) while his sister actually runs the gig.

    • benny_s 23 minutes ago
      Can you elaborate on the Epstein topic? Did I miss something?
    • simonsan 22 minutes ago
      "There's no way I'm going to support a family associated to Epstein with my or my company's money."

      Source?