Recently Yahoo News on-line picked up the following:
OpenAI's GPT-5 is live and free for all after years of
hype by Sam Altman. Here's what it can do.
Business Insider - Lakshmi Varanasi,Brent D. Griffiths,Jyoti
Mann,Robert Scammell
Updated Thu, August 7, 2025 at 11:40 a.m. PDT·3 min read
After weeks of relentless hype, GPT-5 is finally here.
OpenAI officially released the highly anticipated new model on Thursday. It's the latest version of ChatGPT, the company's flagship chatbot, and the most widely used AI model on the market.
OpenAI says the model will be available for free for everyone and that users will no longer need to switch between previous models for different tasks. GPT-5 will switch automatically, the company said, depending on the kind of request and its complexity. GPT-5 will be available in standard, mini, and nano versions. Paid users will have higher usage limits.
Altman said GPT-5 was also the company's fastest model yet.
"One of the things I had been pushing the team on was like, 'Hey, we need to make it way, way faster,'" Altman said. "And I now have this experience of like, 'Are you sure you thought enough?'"
In July, Altman said that "GPT-5 is smarter than us in almost every way." Speaking to podcaster Theo Von, Altman said he asked GPT-5 a question he "didn't quite understand" and it "answered it perfectly," making him feel "useless relative to the AI."
The OpenAI CEO also likened GPT-5's development to the Manhattan Project, the US government's effort to build the atomic bomb during World War II, and said that it made him feel nervous.
"There are moments in the history of science, where you have a group of scientists look at their creation and just say, you know: 'What have we done?'" Altman said.
OpenAI's last notable model launch was GPT-4.5 in February, which Altman described at the time as "the first model that feels like talking to a thoughtful person."
The release of GPT-5 has faced several delays, with the timeline shifting from mid-2024 to mid-2025, to August.
During that time, the number of ChatGPT users has continued to grow. In April, Altman said ChatGPT had 500 million weekly active users. OpenAI said on Thursday that ChatGPT is now on track to hit 700 million weekly active users "this week."
Translation
經過 Sam Altman 多年的炒作宣傳,OpenAI
的 GPT-5 終於正式上線並免費向所有人開放 - 以下是它的功能
經過數週的持續宣傳,GPT-5 終於來了。
OpenAI 於週四正式發布了備受期待的新模型。這是該公司旗艦聊天機器人 ChatGPT 的最新版本,也是市場上最廣泛使用的 AI 模型。
OpenAI 執行長 Sam Altman 在周三與記者的電話會議上稱其為一次「重大升級」和「在通用人工智慧 (AGI) 道路上邁出的重要一步」。他表示,在使用 GPT-5 之後,回到 GPT-4 感覺「很悲」。
OpenAI 表示,該模型將免費向所有人開放,用戶無需再在不同任務之間切換先前的模型。該公司表示,GPT-5 將根據請求的類型及其複雜程度自動切換。 GPT-5 將提供標準版、迷你版和奈米版。付費用戶將有更少的使用限制。
Altman 表示,GPT-5 也是該公司迄今為止速度最快的模型。
Altman 說: 「我一直督促團隊的事情之一是,『嘿,我們需要令它大幅提速』」; 「我現在有類似這樣的經歷:’您確定認為已經夠了嗎?’ 」
團隊表示,GPT-5 將改善對氛圍編碼(矽谷最新熱潮)的功能。
OpenAI 一年多來一直在預告 ChatGPT 的下一代版本,但在最近幾天和幾週內,它的熱度有所提升。週日,Altman 在 X 上分享了 GPT-5 的截圖,要求該模型推薦一部「最發人深省」的 AI 電視節目。
今年 7 月,Altman 表示「GPT-5 幾乎在所有方面都比我們聰明」。Altman 在接受播客 Theo Von 採訪時表示他向 GPT-5 提出了一個 “不太理解”的問題,而 GPT-5 “完美地回答了”,這讓他覺得自己 “在相比 AI 之下一無是處”。
這位 OpenAI 執行長也將 GPT-5 的開發比作曼哈頓計劃(二戰期間美國政府為製造原子彈所做的努力),並表示這讓他感到緊張。
Altman 說: 「在科學史上,總有那麼一刻,一群科學家看著他們的成果,然後問:『我們做了什麼?』」。
OpenAI 上一次值得關注的模型發佈是 2 月份的 GPT-4.5,Altman 當時將其描述為「第一個感覺像在與一位深思熟慮的人交談的模型」。
GPT-5 的發布經歷了幾次延期,時間表從 2024 年中期推遲到
2025 年中期,最終推遲到
8 月。
在此期間,ChatGPT 的用戶數量持續增長。今年 4 月,Altman
表示
ChatGPT 的每周活躍用戶已達 5 億。
OpenAI 週四表示,ChatGPT 預計將在「本週」達到 7 億每週活躍用戶。
So, OpenAI has officially released the highly
anticipated new model. Recently, the
number of ChatGPT users continues to grow. In April, Altman said ChatGPT had
500 million weekly active users. Recently, it is on track to hit 700 million
weekly active users. Apparently, ChatGPT is gaining popularity in these days.
Note:
1. Vibe coding (氛圍編碼)is an
artificial intelligence-assisted software development style popularized by
Andrej Karpathy in February 2025.The term was listed in the Merriam-Webster
Dictionary the following month as a "slang & trending" term. It
describes a chatbot-based approach to creating software where the developer
describes a project or task to a large language model (LLM), which generates
code based on the prompt. Unlike traditional AI-assisted coding or pair
programming, the human developer avoids micromanaging the code, accepts
AI-suggested completions liberally, and focuses more on iterative
experimentation than code correctness or structure. (Wikipedia)
沒有留言:
張貼留言