Claude is unavailable for some users. Web and Mobile. | Hacker News Reader

Claude is unavailable for some users. Web and Mobile.

status.anthropic.com

28 pointscetinsert35 comments

Okay, I'm struck that 90% of the comments in a post about Claude being unavailable are complaints that ChatGPT is unavailable. I guess that probably corresponds to the companies respective share of the market?

Anyway, my tale of woe and recovery re the great centralized chatbot outage was turning to use MLCChat on my phone running a Mistral 7b model and being happily surprised it could actually summarize and answer questions about (though slowly) an article I didn't want to read in its entirety. Although, I didn't quite trust it, so I read the article anyway. The summary wasn't too bad, though. Good to know.

Folks, Claude, ChatGPT, and perplexity all down.-

The singularity is here.-

A coordinated DDoS effort against popular AI front-ends seems like the most likely explanation.

Hanlon's razor: (A) An unimaginably large coordination to attack the most computing power on earth temporarily out of unspecified malice (B) It's finals week and a cascading tidal wave of millions of kids are texting each other that chatgpt is down and they need an alternative

perhaps, but it doesn't offer an explanation for why chatgpt.com is down for this long.

Because they also had an incident overnight, and the resolution caused some cascading issues?

https://status.openai.com/incidents/rs49lmsh6xxw

Or they started querying each other. :)

2024-06-04 12:32 - Perplexity enabled a feature that double-checks responses with Claude

2024-06-04 13:51 - ChatGPT enabled a feature that double-checks responses with Perplexity

2024-06-04 15:04 - Claude enabled a feature that double-checks responses with ChatGPT

2024-06-04 17:42 - The internet ran out of bandwidth as LLMs starts going around in infinitive loops, double-checking double-checked answers

Indeed. But ...

... by whom?

(Cue spooky music)

The api call is coming from inside the server!

Not to buzzkill the humor, but Gemini never went down. Those other 3 did though.

This may be tangential, but situations like this just make me want to try to spin up my own LLM. But I suspect I won't benefit from economics of scale, right?

We run a company LLM, fine-tuned Llama 3. It works well and is dirt cheap to run. We do it for reasons other than availability, though.

Probably OpenAI users flooding other services, unintentional DDOS.

The OpenAI API meanwhile is still up :)

Yep. Weirdly, ChatGPT (non-paying account) is also seemingly hanged.

Paid account is also busted.

This speaks the importance of locally hosted models.

I'd love to use local models, but seems like most of the easy to use software out there (LM Studio, Backyard AI, koboldcpp) doesn't really play all that nicely with my Intel Arc GPU and it's painfully slow on my Ryzen 5 4500. Even my M1 MacBook isn't that fast at generating text with even 7B models.

I wonder if llama.cpp with SYCL could help, will have to try it out: https://github.com/ggerganov/llama.cpp/blob/master/README-sy...

But even if that worked, I'd still have the problem that IDEs and whatever else I have open already eats most of the 32 GB of RAM my desktop PC has. Whereas if I ran a small code model on the MacBook and connected to it through my PC, it'd still probably be too slow for autocomplete, when compared to GitHub Copilot and less accurate than ChatGPT or Phind for most stuff.

Nah, I have my own locally hosted LLMs but Claude is the one I use for long-context stuff. So this situation still impacts me.

"This website is using a security service to protect itself from online attacks. The action you just performed triggered the security solution. There are several actions that could trigger this block including submitting a certain word or phrase, a SQL command or malformed data."

This is what I'm getting from Claude.

I wonder if the OpenAI issue has anything to do with Apple doing some large performance/integration testing ahead of potential WWDC announcement on Monday

HuggingChat is still working though! https://hf.co/chat/

Claude, ChatGPT and perplexityAI are all down.

some people on Twitter also report Gemini being down, perhaps some critical service issue

Mistral is standing still.

Because no one's using Le Chat.

I guess today, at least they got my attention.

So is Gemini

Are ChatGPT and Claude all hosted on the same cloud?

Do they make calls to each other?

Yeah, I just noticed, coincidentally I'm also having issues with ChatGPT, it will not render the chat page as usual...

Ditto. Apparently Gemini has near 100% of the market share right now ;-)

Gemini also unstable.

It's stable for me so far. Are there any problematic prompts?

It's been helping me with some code and I made the prompt below to double check.

A Reddit comment in the style of Edgar Allen Poe:

Good sir (or madam, for the veil of anonymity shrouds us all), I find myself utterly disquieted by the omnipresent algorithm that governs this digital realm. Is it not akin to a raven, perched upon a forgotten server, croaking pronouncements that dictate the content we see?

A chill wind whispers of a bygone era, when the written word flowed freely, unfettered by the unseen hand of calculated manipulation. Now, are we but phantoms in a spectral dance, forever doomed to see only that which the algorithm deems fit for our consumption?

Let us raise a collective glass, not of merriment, but of mournful acceptance, for perhaps, in this digital necropolis, true freedom of expression lies buried six feet deep.

Maybe not a coincidence!

Now we all need to use Google again. (Sigh) What a waste of time. Feels like 2020 again.

Reply on news.ycombinator.com