{"id":32004,"date":"2025-04-01T11:01:29","date_gmt":"2025-04-01T03:01:29","guid":{"rendered":"https:\/\/www.1ai.net\/?p=32004"},"modified":"2025-04-01T11:01:29","modified_gmt":"2025-04-01T03:01:29","slug":"openai-%e5%ae%98%e5%ae%a3%e5%b0%86%e5%8f%91%e5%b8%83%e8%87%aa-gpt-2-%e4%bb%a5%e6%9d%a5%e7%9a%84%e9%a6%96%e4%b8%aa%e5%bc%80%e6%94%be%e6%9d%83%e9%87%8d%e8%af%ad%e8%a8%80%e6%a8%a1%e5%9e%8b%ef%bc%8c","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/32004.html","title":{"rendered":"OpenAI Announces Release of First Open Weighted Language Model Since GPT-2 with Inference"},"content":{"rendered":"

April 1 News.OpenAI<\/a> CEO Sam Altman made an official announcement today that in the coming months he will release a \"Powerful New Open Weighted Language Model<\/strong>\" and has a reasoning function.<\/p>\n

\n
    \n
  • We plan to release the first open weighted language model since GPT-2.<\/li>\n
  • We've been thinking about this for a long time, but other priorities have prevailed. It feels important to do this now.<\/li>\n
  • Prior to release, we will evaluate this model against the prepared framework, just as we would do with any other model. Additional work will also be done given that we know this model will be modified after release.<\/li>\n
  • We still have some decisions to make, so will be hosting developer events to gather feedback and try out early prototypes afterward. We'll be starting in San Francisco in a couple of weeks (with a developer event), followed by sessions in Europe and APAC.<\/li>\n
  • We're looking forward to seeing how developers will build it, and how large corporations and governments will use it where they'd prefer to run the model themselves.<\/li>\n<\/ul>\n<\/blockquote>\n

    \"eaafdc17j00su0qyc00bhd000gi00kfp\"<\/p>\n

    Until now, users have only been able to use their models through OpenAI's interface or by integrating their API keys into their apps, meaning the models themselves remain secret and proprietary.<\/p>\n

    However, open weighting models are different. The pre-trained parameters (the \"weights\" that define how the model responds) in these models are openly shared with the public. Developers and researchers can download them, run them locally, fine-tune them for specific use cases, or plug them into a custom application.IT House warns thatJust because weights are open doesn't mean they are open source<\/strong>, the training code, datasets, and detailed logs may still remain proprietary or unpublished, and thus may not be fully open source.<\/p>\n

    Companies like Meta and Mistral have released powerful models like Llama and Mixtral, and new players like DeepSeek, Alibaba, and Baichuan are launching open source models that can rival them. Many people, from academic researchers to independent developers, are embracing open weighting models, and it's time for OpenAI to make a change.<\/p>","protected":false},"excerpt":{"rendered":"

    On April 1st, OpenAI CEO Sam Altman announced today that he would launch a \u201cstrong new open-weight language model\u201d in the coming months, with reasoning. We plan to publish the first open-weight language model since GPT-2. We have been considering this for a long time, but other priorities have prevailed. Now it feels important to do this. Before we release, we will evaluate the model in the framework we have prepared, as we will do for any other model. Since we know that the model will be revised after its publication, additional work will be done. We still have a number of decisions to take, so a developer event will be held to collect feedback and then try the early prototype. We'll start in San Francisco in a few weeks<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[190],"collection":[],"class_list":["post-32004","post","type-post","status-publish","format-standard","hentry","category-news","tag-openai"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/32004","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=32004"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/32004\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=32004"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=32004"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=32004"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=32004"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}