A postmortem of three recent issues
anthropic.com
A postmortem of three recent issues
1–10 of 120 posts
Re: A postmortem of three recent issues
#2TL;DR — Anthropic Postmortem of Three Recent Issues
In Aug–Sep 2025, Claude users saw degraded output quality due to infrastructure bugs, not intentional changes.
The Three Issues 1. *Context window routing error* - Short-context requests sometimes routed to long-context servers.
- Started small, worsened after load-balancing changes.
- Runtime optimizations wrongly boosted improbable tokens.
- Occasionally dropped the true top token.
- Benchmarks missed these degradations.
- Privacy/safety rules limited access to real user data for debugging.
Fixes and Next Steps - More sensitive, continuous evals on production.
- Better tools to debug user feedback safely.
- Stronger validation of routing, output correctness, and token-selection.
Re: A postmortem of three recent issues
#3TL;DR — Anthropic Postmortem of Three Recent Issues In Aug–Sep 2025, Claude users saw degraded output quality due to infrastructure bugs, not intentional changes. The Three Issues 1. *Context window routing error* - Short-context requests sometimes routed to long-context servers. - Started small, worsened after load-balancing changes. 2. *Output corruption* - TPU misconfigurations led to weird outputs (wrong language…
> Privacy/safety rules limited access to real user data for debugging.
Do their ToS really limit access to user data (prompt/response)? I don't remember seeing anything to that effect in their terms.
Re: A postmortem of three recent issues
#4TL;DR — Anthropic Postmortem of Three Recent Issues In Aug–Sep 2025, Claude users saw degraded output quality due to infrastructure bugs, not intentional changes. The Three Issues 1. *Context window routing error* - Short-context requests sometimes routed to long-context servers. - Started small, worsened after load-balancing changes. 2. *Output corruption* - TPU misconfigurations led to weird outputs (wrong language…
> Privacy/safety rules limited access to real user data for debugging. Do their ToS really limit access to user data (prompt/response)? I don't remember seeing anything to that effect in their terms.
I’d imagine they have a lot of internal controls, even if ultimately someone at the company can read the data within their terms. It makes sense that the teams debugging stuff wouldn’t have this access immediately.
Re: A postmortem of three recent issues
#5Title should be fixed: it’s about Claude models in general, not Claude Code
Re: A postmortem of three recent issues
#6TL;DR — Anthropic Postmortem of Three Recent Issues In Aug–Sep 2025, Claude users saw degraded output quality due to infrastructure bugs, not intentional changes. The Three Issues 1. *Context window routing error* - Short-context requests sometimes routed to long-context servers. - Started small, worsened after load-balancing changes. 2. *Output corruption* - TPU misconfigurations led to weird outputs (wrong language…
> Privacy/safety rules limited access to real user data for debugging. Do their ToS really limit access to user data (prompt/response)? I don't remember seeing anything to that effect in their terms.
I know that when you submit a thumbs up/down rating for a response, you need to opt-in to the whole chat conversation being shared with Anthropic.
Re: A postmortem of three recent issues
#7Wow. Sneaky. They do not even state the rate of impact for the XLA bug afaik, which affected everyone, not just claude code users, very vague. Interesting.
Claude code made almost half a billion so far[1] (>500m in ARR and its like 9 months old) , and 30% of all users have been impacted at least once, just from the first routing bug. Scary stuff.
Their post mortem is basically "evaluations are hard, we relied on vibe checking, now we are going to have even more frequent vibe checking". I believe it was indeed unintentional, but in the future where investor's money wont come down from the skies, serving distilled models will be very tempting. And you can not be liable to any SLA currently, it's just vibes. I wonder how enterprise vendors are going to deal with this going forward, you cannot just degrade quality without client or vendor even being able to really prove it.
[1][https://www.anthropic.com/news/anthropic-raises-series-f-at-...]
Re: A postmortem of three recent issues
#8> Incorrect routing affected less than 0.0004% of requests on Google Cloud's Vertex AI between August 27 and September 16.
Matches my experience. I use CC through our enterprise Vertex AI account and never noticed any degradation.
In general it seems like these bugs, while serious, were substantially less prevalent than anecdotal online reports would have you believe. We are really talking about a ~1-2 week window here where most issues were concentrated, a relatively small percentage of total requests and total users impacted.
Re: A postmortem of three recent issues
#9> We don't typically share this level of technical detail about our infrastructure, but the scope and complexity of these issues justified a more comprehensive explanation.
Layered in aggrandizing. You host a service, people give you money.
Re: A postmortem of three recent issues
#10Seems like Claude is using TPUs a lot more than I thought. For some reason I thought 90%+ of their capacity was from AWS.