Original Source: Xinzhiyuan
Image Source: Generated by Wujie AI
The first OpenAI Developer Conference is an AI feast.
GPT-4 Turbo, significant price reductions, new features for developers, custom GPT, and other major updates have already left AI startups in the dust.
Before people could even digest this, the gravedigger has revealed another bombshell: OpenAI Gobi, also known as GPT-5 multimodal model, will be released in early 2024.
Currently, there are over 160,000 people watching online.
According to Roemmele, Gobi is currently being trained on a massive dataset. It will not only support text and images, but also videos.
Someone asked, "OpenAI insiders claim that the next generation model has achieved true AGI. Have you heard about this?"
The whistleblower said, "GPT-5 is already capable of self-correction and has a certain level of self-awareness. People I know have already seen its demonstration, and currently, seven government agencies are testing the latest model."
Remember, Sam Altman hinted at the end of the conference that OpenAI is embarking on the next major innovation, and everyone will soon realize that what was released today is not worth mentioning.
This aligns perfectly with the information revealed by the whistleblower.
Prior to the developer conference, Roemmele correctly leaked the customizable "GPTs," further confirming the credibility of the GPT-5 rumors.
Just now, Roemmele also revealed that one of OpenAI's strongest competitors, Anthropic, which launched the Claude model, is also about to announce a major news.
Is GPT-5 Really Coming?
Altman has always remained tight-lipped about the next generation of more advanced models.
In June, Altman stated that GPT-5 still has a long way to go before it's ready for training, and there is still a lot of work to be done. He added that OpenAI is exploring new ideas, but they are not yet ready to start researching GPT-5.
Even Microsoft co-founder Bill Gates predicted that GPT-5 would not offer significant performance improvements over GPT-4.
However, while various media leaks may not be entirely reliable, they cannot be completely dismissed.
In fact, as early as April, someone had revealed that OpenAI was internally training a multimodal model called Gobi.
This model has a context window of 64k, but due to the enormous computational power required, it cannot be released at the moment.
In early September, Mustafa Suleyman, co-founder of DeepMind and current CEO of Inflection AI, dropped a bombshell in an interview - he speculated that OpenAI is secretly training GPT-5.
Suleyman believes that Sam Altman recently said they are not training GPT-5, which may not be entirely true. (His exact words were: "Come on. I don’t know. I think it’s better that we’re all just straight about it.")
In the same month, The Information leaked that a brand new multimodal large model named Gobi is being prepared in secret.
Following GPT-4 Vision, OpenAI may introduce a more powerful multimodal large model, codenamed Gobi.
Unlike GPT-4, Gobi is being built as a multimodal model from the start.
Another netizen came out to say that OpenAI originally planned to release the GPT-4 V model at the first Developer Day, but due to the leaks of Gobi and Arrakis, they had to release it early.
This flaw can be seen from the references in the GPT-4 V technical report.
Now, whether Gobi is GPT-5 or not, from the information leaked from various sources, it is currently one of the projects that the OpenAI team is working on.
In late September, Roemmele revealed that Gemini, which he visited at Google, can rival GPT-4.
If the rumors about GPT-5 are true, Google will have to use Gemini to compete with the more advanced GPT-5.
At the conference, a series of updates from OpenAI once again proved its leadership in the field of generative AI to the world.
If competitors like Google want to catch up with the success paradigm of ChatGPT and API services, they still have a lot of work to do.
From the popularity of this conference, it is clear that OpenAI has no direct competitors.
ChatGPT Crashes, Netizens React
Just last night, many people found that their ChatGPT suddenly went offline, sparking public outcry.
Shortly after, OpenAI announced that both the ChatGPT server and the API had crashed for a full 2 hours due to overwhelming demand.
In response, Altman himself immediately made a rather "Versailles" response:
The usage of the new features we released at the developer conference far exceeded our expectations.
We had planned to open GPT to all subscribed users on Monday, but we still couldn't do it. We hope to be able to do it soon. Due to the high load, the service may be unstable in the short term. We apologize for this.
The updated announcement indicates that ChatGPT and the API will still experience some intermittent crashes.
From this graph, you can see the time period of the crash.
For people around the world, the ChatGPT crash is simply a huge disaster.
Without ChatGPT, everyone's work status looks like this....
When the world needs ChatGPT the most, it disappears.
How is your ChatGPT doing?
Reference:
https://the-decoder.com/openais-altman-says-todays-ai-will-be-quaint-by-2024-in-line-with-first-gpt-5-rumor/
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
