Image source: Visual China
Recently, OpenAI has tightened regional restrictions, which has aroused great concern in the tech circle.
On June 25, multiple developers said on social media that they had received a "warning letter" from OpenAI, in which they said that additional measures would be taken from July 9 to block API traffic from unsupported countries and regions.
This isn't the first time OpenAI has set limits based on regional scope. The list given by OpenAI's official website shows that its API service currently supports access to 188 countries and regions, excluding Chinese mainland and Hong Kong, China.
Although OpenAI has not opened its services to the Chinese market before, this "suspension" may mean that the industry leader, which has swept the world with ChatGPT, will successively close various channels that were previously missed, restricting individual users from the C-end to developers and startups on the B-side.
In addition to the APIs officially provided by OpenAI, Chinese enterprise users can still choose Microsoft's Azure OpenAI service to use the interface. On June 27, Microsoft said in a statement that there were "no changes" to the Azure OpenAI service offered in Hong Kong. Microsoft added that "as an independent company, OpenAI will make its own decisions. ”
Chinese large-scale model manufacturers, which have been fighting in the field of artificial intelligence for a long time, have responded quickly to this, trying to win more users who will "move" from OpenAI.
On June 25, Baidu Intelligent Cloud Qianfan launched a large model inclusive plan to provide newly registered enterprise users with services such as 0 yuan call, 0 yuan training, and 0 yuan migration. Among them, the Wenxin flagship model is free for the first time, and an additional ERNIE3.5 flagship model Tokens package with the same scale as OpenAI will be given to OpenAI migration users.
Alibaba, while choosing to carry out the price war to the end, also showed the signboard of "the world's top five, performance comparable to GPT-4". Alibaba Cloud announced that it will provide OpenAI API users with a cost-effective alternative to China's large models, as well as 22 million free tokens and exclusive migration services for Chinese developers. The API price of its qwen-plus model, which has reached the "GPT-4 level", is only 1/50 of OpenAI.
Tencent Cloud also announced that by 24:00 on July 31, 2024, it will give away 100 million Tencent Hybrid Model tokens for free to newly migrated enterprise users, and provide free exclusive migration tools and services. On the 26th, ByteDance released an intelligent development tool based on the large model of Doubao - Doubao MarsCode, which is free and open to Chinese developers.
The dark side of the moon, which ranks in the first echelon of large-scale model startups, said that the dark side of the moon has been compatible with OpenAI's API from the beginning, and developers can directly replace it with the API key of the dark side of the moon. With its long text capability, the company's Kimi intelligent assistant has become a star product in China's AI large model track.
In addition to the manufacturers who want to catch this wave of traffic, the heat has also spread to the A-share market. On June 26, after a number of Chinese large-scale model manufacturers released corresponding "moving" plans, the concept of AI application rose, led by corpus, Sora, and multimodality.
The model side and the application side provide opportunities for technology giants and AI entrepreneurs in different directions. When OpenAI becomes "Close AI" and implements stricter restrictions, developers who rely on OpenAI's APIs still have alternatives to choose from, and startups that shell OpenAI may usher in a "catastrophe" and be the first to be out of this reshuffle. The vacated market share will flow to manufacturers with independent R&D capabilities, which is a test and an excellent opportunity for Chinese large-scale model manufacturers.
International rivals of Chinese tech giants are also likely to benefit from the incident, including Amazon's investment in AI unicorn Anthropic, Gemini's Google, which wants to compete with OpenAI, and Meta Llama 3, a widely influential open-source model. However, the data compliance and security issues behind OpenAI's "drawing the line" may also be unavoidable for Chinese developers, and many Silicon Valley giants may not be able to avoid potential risks.
In the new round of competition for B-end users, Chinese manufacturers need to prove their strength. On the whole, there are no products that have "one ride" and widen the strength gap between China's large manufacturers with independent research and development capabilities, and they are highly substitutable. In terms of inference capabilities and performance, it is still unknown whether it can meet the needs of original OpenAI users and achieve steady growth.
When OpenAI is completely absent, the industry benchmark established by competition from local Chinese manufacturers also needs to be revisited.
In April, Tsai Chongxin, chairman and co-founder of Alibaba, said in an interview that Chinese companies were under competitive pressure in the field of artificial intelligence development, "I think China is lagging behind now." It's clear that an American company like OpenAI is really ahead of all the others. But China is trying to catch up. Tsai believes that other countries are also in the process of rapid development, and China may be two years behind the top large language models in the field of AI technology.
In the context of the global technology competition, it may be called fortunate that OpenAI cannot escape the "window period" of the explosion.
Recently, OpenAI CTO Mira Murati revealed in an interview that GPT-5 may not appear until the end of 2025 to the beginning of 2026. In addition, the release of the GPT-4o-based ChatGPT voice assistant feature will also be delayed. OpenAI said the exact timeline depends on meeting "high security and reliability standards."
In May this year, OpenAI co-founder and chief scientist Ilya Sutskovo officially announced his resignation, putting an end to the infighting. At the same time, OpenAI's team responsible for human security was revealed to be disbanded. As OpenAI begins training the next generation of cutting-edge models, the company's course also leaves new suspense on user expectations and the boundaries of general AI.
Before the emergence of new "killer" products, Chinese manufacturers were making a decision for developers to roll their own models into AI applications. This ecological battle will not end with a successful "move" from OpenAI. (Fortune Chinese Network)
At Fortune's Brainstorm AI conference in Singapore on July 30-31, 2024, guests will discuss the progress, applications, and impact of AI technologies. For more information, check out the poster below.
At Fortune Plus, netizens have made many in-depth and thoughtful comments on this article. Let's take a look. You are also welcome to join us and talk about your ideas. Other hot topics today:
Check out the great view of "Masayoshi Son: Will Build a Super AI with 10,000 Times More Capabilities than Humans".
Check out the wonderful views of "Ministry of Finance: Stamp duty on securities transactions from January to May was 43.9 billion yuan, down 50.8% year-on-year".