{"id":29787,"date":"2025-02-28T10:31:45","date_gmt":"2025-02-28T02:31:45","guid":{"rendered":"https:\/\/www.1ai.net\/?p=29787"},"modified":"2025-02-28T10:31:45","modified_gmt":"2025-02-28T02:31:45","slug":"%e5%a5%a5%e5%b0%94%e7%89%b9%e6%9b%bc%e5%9d%a6%e8%af%9a-openai-%e6%80%a5%e7%bc%ba-gpu%ef%bc%8cgpt-4-5-%e5%8f%aa%e8%83%bd%e5%88%86%e9%98%b6%e6%ae%b5%e6%8e%a8%e5%87%ba","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/29787.html","title":{"rendered":"Altman admits OpenAI is short of GPUs, GPT-4.5 can only be rolled out in phases"},"content":{"rendered":"<p>February 28th.<a href=\"https:\/\/www.1ai.net\/en\/tag\/openai\" title=\"[View articles tagged with [OpenAI]]\" target=\"_blank\" >OpenAI<\/a> Chief Executive Officer<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e5%a5%a5%e5%b0%94%e7%89%b9%e6%9b%bc\" title=\"[Sees articles with [Altman] labels]\" target=\"_blank\" >Oltmann (name)<\/a>said today that as a result of the \"<a href=\"https:\/\/www.1ai.net\/en\/tag\/gpu\" title=\"_OTHER ORGANISER\" target=\"_blank\" >GPU<\/a> resource constraints,\" the company had to<strong>phased roll-out<\/strong>updated <a href=\"https:\/\/www.1ai.net\/en\/tag\/gpt-4-5\" title=\"[SEE ARTICLES CONTAINING [GPT-4.5] LABELS]\" target=\"_blank\" >GPT-4.5<\/a> Model.<\/p>\n<p>In a post on the X platform, Altman noted that GPT-4.5 was a \"large\" and \"expensive\" model that would require<strong>Tens of thousands of additional GPUs<\/strong>\u00a0GPT-4.5 will be rolled out to ChatGPT Pro subscribers first, followed by ChatGPT Plus users next week.<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-29788\" title=\"a73a6c81j00ssdg9r0033d000hw00oup\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2025\/02\/a73a6c81j00ssdg9r0033d000hw00oup.jpg\" alt=\"a73a6c81j00ssdg9r0033d000hw00oup\" width=\"644\" height=\"894\" \/><\/p>\n<p>The high cost of GPT-4.5 stems in part from its enormous scale. openAI charges $75 per million input tokens (note: currently around Rs. 546) and $150 per million generated tokens. That's 30 times the input cost and 15 times the output cost of its main model, GPT-4o.<\/p>\n<p>\"We've been growing rapidly.<strong>But now the GPU is not enough.<\/strong>.\" Altman writes, \"We will be adding next week<strong>Tens of thousands of GPUs<\/strong>, and rolled out ...... to Plus users at that time It's not our ideal modus operandi, but it's hard to accurately predict the growth fluctuations that lead to GPU shortages.\"<\/p>\n<p>Altman has said that a lack of computing power is causing delays in the company's product releases. openAI plans to address this issue in the next few years by developing its own AI chips and building a vast network of data centers.<\/p>","protected":false},"excerpt":{"rendered":"<p>Feb. 28, 2012 - OpenAI CEO Michael Altman said today that the company had to roll out its latest GPT-4.5 model in stages due to a \"shortage of GPU resources. In a post on the X platform, Altman noted that GPT-4.5 is a \"huge\" and \"expensive\" model, requiring tens of thousands of additional GPUs to make it available to more ChatGPT users. GPT-4.5 will be available to ChatGPT Pro subscribers first, followed by ChatGPT Plus users next week. The high cost of GPT-4.5 stems in part from its sheer size; OpenAI charges $75 per million input tokens ($10 per million input tokens).<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[1669,415,190,5731],"collection":[],"class_list":["post-29787","post","type-post","status-publish","format-standard","hentry","category-news","tag-gpt-4-5","tag-gpu","tag-openai","tag-5731"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/29787","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=29787"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/29787\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=29787"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=29787"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=29787"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=29787"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}