← sources

I don't believe Claude is nerfing models - it's simply giving higher compute priority to better quality projects.

source: reddit (reddit:ClaudeAI)published 2026-10-03 16:04ingested 2026-10-04 02:35

retained text

I'm going to say off the bat that there's no way for me to quantitatively measure this by any means possible, but I do believe that routing on Claude's backend is taking several factors in to consideration and I wouldn't be at all surprised if it's based on your project quality and personality.  I haven't seen any drops in performance since Sonnet 5.5 and Opus 5.5 have come out beyond my weekly usage being eaten up quickly. I'm on the $200/mo plan and tend to have multiple lanes doing work concurrently with one orchestration layer and this workflow has been incredibly competent for me.

Maybe it's because I'm following a set of evolving best practices I glean from this sub and other parts of the web, but I do not believe that there's the sort of Raising Cain's model of releasing high quality then dropping quality later on for every user as popularity and usage goes.  For those who don't know, Raising Cains is a chicken place that, when they open a new location, draws in new customers with larger pieces of chicken.  As popularity increases, they decrease the size of their chicken strips to deal with growing demand at that location.  

Anyway, I do believe that - based purely on observations of people's posted interactions with Claude in this sub and others - that Anthropic routes your project work based on how you've interacted with Claude.  Are you acting like a seasoned developer and PM? Are you cursing or using "bro" language with it? Are you actually giving it proper instructions or just asking it to one-shot without any guidance? My thought is that if you're giving it low-quality inputs, then expect the processing and output to match that same energy.  

And to me this makes sense from a business standpoint as well.  Their system is reading every character of every message and piece of code it's writing.  Why wouldn't it make a determination to treat some projects better than others? Some animals are more equal than others, and if a project that they deem as having a higher chance of being successful is going to keep that user within this ecosystem why not treat it as such? Why would they give lower quality projects the same level of processing power as something that clearly has the ability to scale?  If all you're doing is building a game in browser or another usage dashboard, would you as Anthropic give that the same priority as a saas that's confirmed to have customers and is scaling? especially when that saas now becomes a better candidate for an enterprise plan and more usage? 

These guys know who's getting to market and finding traction and who isn't.  It makes zero sense from a consumption standpoint to put everyone on equal footing and they have the tools to determine who's worth helping and who can deal with lower compute power while still paying a subscription.  

document fetched 2026-10-04 00:21 · canonical https://reddit.com/r/ClaudeAI/comments/1wwqv7t/i_dont_believe_claude_is_nerfing_models_its