OpenAI skilled a partial outage on Tuesday morning that created some points for customers attempting to entry ChatGPT, Sora, and the API, the corporate mentioned on its standing web page.
The corporate began investigating the problems late on Monday night time, however the partial outage has persevered by way of Tuesday morning. Round 5:30 a.m. PT on Tuesday, OpenAI says it recognized the difficulty and began working to repair it.
Nonetheless, at roughly 8 a.m. PT, OpenAI mentioned full restoration throughout its companies might take “another few hours,” which means of us logging onto work on the West Coast of the US gained’t have the ability to entry ChatGPT this morning.
Whereas ChatGPT outages sometimes final just some hours, Tuesday’s partial outage is notably lengthy in length. The corporate mentioned customers might expertise “elevated errors and latency” when utilizing ChatGPT.
When TechCrunch tried to entry GPT-4o in ChatGPT on Tuesday morning, the chatbot responded with an error studying “Too many concurrent requests.”
Tuesday’s partial outage comes amid a flurry of bulletins from OpenAI. At Apple’s WWDC occasion on Monday, the iPhone maker introduced deeper integrations with OpenAI’s fashions. Additionally on Monday, an OpenAI spokesperson confirmed to CNBC that the corporate has reached $10 billion in annualized recurring income. Earlier on Tuesday, OpenAI CEO Sam Altman introduced an 80% value minimize for builders attempting to entry its o3 AI reasoning fashions in its API.
OpenAI has beforehand run into points scaling the utilization of its AI fashions to a whole bunch of tens of millions of customers, however that’s precisely what the corporate must do to satisfy its grand ambitions. Altman has beforehand mentioned the corporate’s “GPUs are melting” to maintain up with the demand for ChatGPT, indicating that OpenAI’s computing assets are unfold skinny lately. Evidently demand for OpenAI’s fashions is simply persevering with to rise.
