It’s a new round of instability for the chatbot. The company has put out word that they are looking into an "elevated error rate across multiple models.” It’s the sort of wobble that makes the competition for dependability all the more pressing.
If you look at the outage trackers, this is no isolated case. We’re seeing reports of things not working in chat, for coding or over the API, and it seems to come down to where you are and what you’re trying to do. For some, it’s not so much a total blackout as an on-again, off-again affair, which can make for an unpredictable workday.
What users are seeing today
People are telling us their responses won’t load, a conversation will just die in the middle of it, or a code session will be dropped. Some developers are running up against server-side errors, which is a sign of strain on the back end, not your internet connection. It’s been a mixed bag on both web and mobile, so the problem is everywhere you look.
Here is what we are hearing from forums and the like:
– API Error: 500 Internal Server Error
– The chatbot gives you nothing, or a blank screen
– Requests are either slow or don’t go through
– Claude Code gets cut off
Outage metrics from Downdetector
Downdetector has seen a run of complaints well into the evening of the 23rd. When it was at its worst, they had 6,740 on the books, up from a typical 11. You could call it a service-wide issue; it’s not just one odd feature or a few towns in a corner of the world.
Anthropic’s response and current status
Over on the community channels, Anthropic has confirmed the high error rates and says they are on it. We are still waiting for a full statement, and they haven’t put a finger on the cause or when it will be sorted out.
As of now, you can still get to Claude in a way. Some of your requests will work, others won’t, and it depends on if you are using the API, the chat or Claude Code. It’s a pattern we’ve seen in recent months when demand has outstripped what they can put in front of it.
You see it in the numbers: 500-series responses, model-level hiccups, and the kind of stress that comes with a lot of people piling in. It’s a scaling problem that has a way of rearing its head as more of the world starts to use Claude.
Why this outage matters in the AI race
When you are building for enterprise, how often something is up is just as good as how smart it is. If you can’t count on it, you put off shipping code and you start to look at other options. In a field with ChatGPT, Gemini, Copilot and Perplexity, being reliable is what sets you apart.
The truth of the matter is that with AI woven into our writing, our research and our to-do lists, an hour of downtime is a lot to sit with. You can see it in the way the complaints pile up the moment a top-tier platform can’t handle the load.
What you can do right now
If you are having a rough time with Claude, here is a way to make things a bit easier:
– Give it a minute and try again
– Do a hard refresh or start a new session
– See what Anthropic has to say on their status page
– Move to a different model for the time being
For the developers out there, some retry logic with a backoff will help keep things moving. And if you need to get on with it, there is always OpenAI, Google, Microsoft or Perplexity while you wait for word from Anthropic.
Now we will have to see how fast they can put the brakes on this. With the volume of reports and the services involved on this date, it will be about more than just the fix – it will be about whether they can put these scaling headaches to rest for good.
