OpenAI tackles 'major outage' hitting ChatGPT APIs
Meltdown apparently resolved, capacity issues still popping up, as Claude hits a resource wall
Final update ChatGPT's hundreds of millions of users were left without their digital discussion companion this morning, as owner OpenAI reported a major outage across the platform and its APIs.
Open AI identified the problem at 0554 PT (1354 UTC), indicating an hour later that it had "identified an issue resulting in high error rates across the API and ChatGPT," and that it was working on remediation.
As of 0733 PT (1533 UTC), the company said a fix had been implemented and services were beginning to recover. Fifteen minutes later, OpenAI said it was "now seeing normal responses from our services."
The incident comes just days after OpenAI held its first developer conference in San Francisco, where it announced a new model called GPT-4 Turbo, which can accept verbose prompts of up to 128,000 tokens, or roughly 300 pages of text. The new model will also be updated to include information dated up to April 2023, as opposed to the September 2021 knowledge cap the older models faced.
"We will try to never let it get that out of date again," OpenAI CEO Sam Altman said.
- OpenAI could be valued up to $90 billion if deal to sell employee shares closes
- OpenAI warns folks over GPT-4 Vision's limits and flaws
- Bad Vibrations: Music publishers sue Anthropic AI for using copyrighted lyrics
- ChatGPT study suggests its LLMs are getting dumber at some tasks
As to the nature of today's outage, that's anyone's guess, and OpenAI didn't respond to our questions. The service still isn't completely steady, however. "We're experiencing exceptionally high demand. Please hang tight as we work on scaling our systems," a popup on ChatGPT told us when we visited.
Also worth noting is that Claude – an AI bot built by Anthropic, which was set up by some ex-OpenAI staff in 2021 and has since seen heavy investment from Amazon – was unavailable for some folks.
"Due to unexpected capacity constraints, Claude is unable to respond to your message," the bot said today when this reporter asked if it was feeling okay. We didn't hear back from Anthropic about this bother this morning, either.
We will update this story if we learn more. ®
Updated to add
Claude may not be out for the count. "Due to heightened demand we have made adjustments to prioritize capacity but can confirm we have not seen an outage," an Anthropic spokesperson told The Register.
Meanwhile, OpenAI says the interruptions to service were caused by loading issues.
OpenAI not only suffered technical issues, it also seems to have been DDoSed.