今日，网络盛传OpenAI即将推出的GPT-4.5 Turbo模型信息在多个搜索引擎上意外曝光，引发广泛关注。据网友发现，该模型同时出现在Bing、谷歌和DuckDuck Go的搜索结果中，介绍称其为OpenAI迄今为止速度最快、准确度最高且可扩展性最强的AI模型。上下文窗口高达256K，是GPT-4的两倍。更有网友推测，官方发布会可能选在本周四，即GPT-4发布一周年纪念日。此推测或因OpenAI首席执行官Sam Altman即将参加Lex Fridman的播客节目而得到印证。预告信息还显示，GPT-4.5 Turbo的“知识更新截止日期”为2024年6月，暗示原定于6月发布的模型可能已被提前泄露。
英文标题：OpenAI GPT-4.5 Turbo Model Leaked Ahead of Schedule, Possible Launch on Thursday
英文关键词：AI model, GPT-4.5 Turbo, Early Leak
英文新闻内容：
Today, reports emerged that information regarding the highly anticipated GPT-4.5 Turbo model from OpenAI has been accidentally exposed across multiple search engines, triggering widespread excitement. According to online users, the model appeared in search results from Bing, Google, and DuckDuck Go, with the description highlighting it as the fastest, most accurate, and most scalable AI model from OpenAI to date. The context window of GPT-4.5 Turbo is said to be as high as 256K, double that of GPT-4. There is also speculation that the official launch might be scheduled for Thursday, coinciding with the one-year anniversary of GPT-4’s release. This speculation gains weight from the fact that OpenAI CEO Sam Altman is set to appear on Lex Fridman’s podcast, which he also did following the GPT-4 launch last year. The teaser information also reveals that the “knowledge update cutoff date” for GPT-4.5 Turbo is June 2024, suggesting that the model, originally slated for a June release, might have been leaked ahead of schedule.
【来源】https://mp.weixin.qq.com/s/hfOClnutRNguxmFdjkggvA
Views: 1