{"id":126183,"date":"2025-05-23T20:37:13","date_gmt":"2025-05-23T20:37:13","guid":{"rendered":"https:\/\/www.europesays.com\/uk\/126183\/"},"modified":"2025-05-23T20:37:13","modified_gmt":"2025-05-23T20:37:13","slug":"openai-upgrades-the-ai-model-powering-its-operator-agent","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/uk\/126183\/","title":{"rendered":"OpenAI upgrades the AI model powering its Operator agent"},"content":{"rendered":"<p id=\"speakable-summary\" class=\"wp-block-paragraph\">OpenAI is updating the AI model powering\u00a0<a href=\"https:\/\/techcrunch.com\/2025\/01\/23\/openai-launches-operator-an-ai-agent-that-performs-tasks-autonomously\/\" target=\"_blank\" rel=\"noreferrer noopener\">Operator<\/a>, its AI agent that can autonomously browse the web and use certain software within a cloud-hosted virtual machine to fulfill users\u2019 requests.<\/p>\n<p class=\"wp-block-paragraph\">Soon, Operator will use a model based on <a href=\"https:\/\/techcrunch.com\/2025\/04\/02\/openais-o3-model-might-be-costlier-to-run-than-originally-estimated\/\" target=\"_blank\" rel=\"noopener\">o3<\/a>, one of the latest in OpenAI\u2019s o series of \u201creasoning\u201d models. Previously, Operator relied on a custom version of <a href=\"https:\/\/techcrunch.com\/2024\/05\/13\/openais-newest-model-is-gpt-4o\/\" target=\"_blank\" rel=\"noopener\">GPT-4o<\/a>. <\/p>\n<p class=\"wp-block-paragraph\">By many benchmarks, o3 is a far more advanced model, particularly on tasks involving math and reasoning. <\/p>\n<p class=\"wp-block-paragraph\">\u201cWe are replacing the existing GPT\u20114o-based model for Operator with a version based on OpenAI o3,\u201d OpenAI <a href=\"https:\/\/openai.com\/index\/o3-o4-mini-system-card-addendum-operator-o3\/\" target=\"_blank\" rel=\"noreferrer noopener nofollow\">wrote<\/a> in a blog post. \u201cThe API version [of Operator] will remain based on 4o.\u201d<\/p>\n<p class=\"wp-block-paragraph\">Operator is one among many agentic tools released by AI companies in recent months. Companies are racing to make highly sophisticated agents that can reliably carry out chores more or less without supervision.<\/p>\n<p class=\"wp-block-paragraph\">Google offers a \u201c<a href=\"https:\/\/techcrunch.com\/snippet\/3009950\/new-dev-tools-round-out-the-first-day-of-google-i-o\/\" target=\"_blank\" rel=\"noopener\">computer use<\/a>\u201d agent through its Gemini API that can similarly browse the web and take actions on behalf of users, as well as a more consumer-focused offering called <a href=\"https:\/\/techcrunch.com\/2025\/05\/20\/google-rolls-out-project-mariner-its-web-browsing-ai-agent\/\" target=\"_blank\" rel=\"noopener\">Mariner<\/a>. Anthropic\u2019s models are also able to <a href=\"https:\/\/techcrunch.com\/2024\/10\/22\/anthropics-new-ai-can-control-your-pc\/\" target=\"_blank\" rel=\"noopener\">perform computer tasks<\/a>, including opening files and navigating web pages.<\/p>\n<p class=\"wp-block-paragraph\">According to OpenAI, the new Operator model, called o3 Operator, was \u201cfine-tuned with additional safety data for computer use,\u201d including datasets designed to \u201cteach the model [OpenAI\u2019s] decision boundaries on confirmations and refusals.\u201d<\/p>\n<p class=\"wp-block-paragraph\">OpenAI has released a technical report showing o3 Operator\u2019s performance on specific safety evaluations. Compared to the GPT-4o Operator model, o3 Operator is less likely to refuse to perform \u201cillicit\u201d activities and search for sensitive personal data, and less susceptible to a form of AI attack known as prompt injection, per the technical report.<\/p>\n<p class=\"wp-block-paragraph\">\u201co3 Operator uses the same multi-layered approach to safety that we used for the 4o version of Operator,\u201d OpenAI wrote in its blog post. \u201cAlthough o3 Operator inherits o3\u2019s coding capabilities, it does not have native access to a coding environment or terminal.\u201d<\/p>\n","protected":false},"excerpt":{"rendered":"OpenAI is updating the AI model powering\u00a0Operator, its AI agent that can autonomously browse the web and use&hellip;\n","protected":false},"author":2,"featured_media":34000,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3163],"tags":[323,1942,1318,55857,53,16,15],"class_list":{"0":"post-126183","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-artificial-intelligence","8":"tag-ai","9":"tag-artificial-intelligence","10":"tag-openai","11":"tag-operator","12":"tag-technology","13":"tag-uk","14":"tag-united-kingdom"},"share_on_mastodon":{"url":"https:\/\/pubeurope.com\/@uk\/114559073822630087","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/posts\/126183","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/comments?post=126183"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/posts\/126183\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/media\/34000"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/media?parent=126183"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/categories?post=126183"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/tags?post=126183"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}