{"id":52165,"date":"2023-03-14T19:12:36","date_gmt":"2023-03-14T19:12:36","guid":{"rendered":"https:\/\/mailinvest.blog\/index.php\/2023\/03\/14\/openai-just-upgraded-chatgpt-with-the-new-gpt-4-model\/"},"modified":"2023-03-14T19:13:55","modified_gmt":"2023-03-14T19:13:55","slug":"openai-just-upgraded-chatgpt-with-the-new-gpt-4-model","status":"publish","type":"post","link":"https:\/\/mailinvest.blog\/index.php\/2023\/03\/14\/openai-just-upgraded-chatgpt-with-the-new-gpt-4-model\/","title":{"rendered":"OpenAI just upgraded ChatGPT with the new GPT-4 model"},"content":{"rendered":"

<\/a>\n<\/p>\n

\n
\n

GPT-4 is out<\/p>\n


\n \"\"\/
\n <\/a>\n <\/div>\n

\n


\n
\n AppleInsider might earn an affiliate fee on purchases made by hyperlinks on our website.
\n <\/em> \n <\/p>\n<\/div>\n

OpenAI has launched GPT-4, an upgraded language mannequin for ChatGPT that’s extra correct \u2014 however nonetheless flawed. <\/p>\n

\n

ChatGPT is a chatbot that anyone can use<\/a> to generate tales, articles, and different types of textual content. The long-awaited GPT-4 replace can settle for pictures as inputs however nonetheless generates textual content for the output. <\/p>\n<\/div>\n

\n

OpenAI calls it “the most recent milestone in its effort in scaling up deep studying” in its Tuesday announcement<\/a> and claims it has proven human-level efficiency on varied skilled and tutorial benchmarks. For instance, it handed a simulated bar examination with a rating across the high 10% of check takers. <\/p>\n<\/div>\n

\n

The corporate has rebuilt its deep studying stack over the previous two years and labored with Microsoft Azure to design a supercomputer for workloads. GPT-3.5 was the primary check run of the brand new system. <\/p>\n<\/div>\n

\n

Whereas the brand new system is extra correct than its predecessor, it nonetheless makes factual errors. For instance, in one of many examples that OpenAI offers it referred to Elvis Presley because the son of an actor. <\/p>\n<\/div>\n

\n
\"\"<\/a>\n<\/div>\n<\/div>\n
\n

GPT-4 additionally lacks data of occasions after September 2021 and might’t be taught from its experiences. Nevertheless, OpenAI has improved lots of GPT-4’s security properties. <\/p>\n<\/div>\n

\n

The corporate decreased the mannequin’s tendency to reply to requests for disallowed content material by 82% in comparison with GPT-3.5. And it responds to delicate requests similar to medical recommendation and self-harm per the corporate’s insurance policies 29% extra usually. <\/p>\n<\/div>\n

\n

“We stay up for GPT-4 changing into a helpful device in bettering folks’s lives by powering many functions,” OpenAI wrote. “There’s nonetheless numerous work to do, and we stay up for bettering this mannequin by the collective efforts of the neighborhood constructing on high of, exploring, and contributing to the mannequin.” <\/p>\n<\/div>\n

\n

OpenAI is releasing GPT-4’s textual content enter functionality into ChatGPT and its API with a waitlist<\/a>. Nevertheless, it is instantly obtainable as we speak for paying customers with the ChatGPT Plus<\/a> $20 month-to-month subscription plan. <\/p>\n<\/div>\n

\n

Folks can even expertise ChatGPT and GPT-4 in Microsoft Bing and the Edge browser, which the corporate introduced in February<\/a>.<\/p>\n<\/div><\/div>\n