据知情人士透露，OpenAI正筹备推出一系列全新的人工智能模型，其中包括一个可能被命名为GPT-4.1的增强型多模态版本。据悉，这款模型是去年发布的GPT-4o的升级版，能够实时处理音频、图像和文本等多种信息。
除了GPT-4.1之外，OpenAI还计划发布其简化版本，即GPT-4.1 mini和nano。有消息称，这些新模型可能会在下周正式亮相。
与此同时，OpenAI也在准备推出其o3推理模型的完整版以及o4 mini版本。近期，AI工程师Tibor Blaho在ChatGPT的新网页版中发现了对o4 mini、o4 mini high及o3的引用，这似乎暗示相关功能即将上线。根据目前的消息，o3和o4 mini预计将在下周发布，但具体时间可能因公司调整而有所变化。
此前，OpenAI首席执行官Sam Altman曾在社交平台X上预告称，公司将推出一项令人期待的新功能。然而，尚不清楚这一功能是否与ChatGPT中提到的o3或o4 mini版本有关。
值得一提的是，由于容量问题，OpenAI近期多次推迟了新模型的发布时间。Altman本月早些时候曾表示，由于面临资源限制，客户需要为新版发布延迟、潜在的功能故障以及服务响应速度下降做好心理准备。上个月，OpenAI的高级图像生成功能因需求过高而不得不限制请求量，Altman甚至开玩笑称“我们的GPU都快烧毁了”。
截至目前，OpenAI尚未对上述消息作出官方回应。
According to insiders, OpenAI is preparing to launch a series of new artificial intelligence models, including an enhanced multimodal version that may be named GPT-4.1. This model is said to be an upgraded version of GPT-4o, which was released last year and can process audio, images, and text in real time.
In addition to GPT-4.1, OpenAI also plans to release its streamlined versions, namely GPT-4.1 mini and nano. Reports suggest these new models could debut as early as next week.
At the same time, OpenAI is also gearing up to roll out the full version of its o3 inference model as well as the o4 mini version. Recently, AI engineer Tibor Blaho discovered references to o4 mini, o4 mini high, and o3 in the new web version of ChatGPT, hinting that these features are about to go live. Based on current information, both o3 and o4 mini are expected to launch next week, although the exact timing may vary due to company adjustments.
Previously, OpenAI CEO Sam Altman hinted on social platform X that the company would introduce an exciting new feature. However, it remains unclear whether this feature is related to the o3 or o4 mini versions mentioned in ChatGPT.
Notably, due to capacity issues, OpenAI has repeatedly delayed the release of new models in recent times. Earlier this month, Altman stated that customers should prepare for potential delays in new releases, possible functionality glitches, and slower service responses due to resource constraints. Last month, OpenAI even had to limit requests for its advanced image generation feature due to overwhelming demand, with Altman joking that "our GPUs are almost burning out."
As of now, OpenAI has not provided an official response to the above reports.
