"ChatGPT+" ecological approach? OpenAI may launch an AI model store

Source: "Science and Technology Innovation Board Daily"

Author: Zheng Yuanfang

Image source: Generated by Unbounded AI tool

After the "iPhone moment", OpenAI may follow Apple's example and use the App Store to create its own application ecosystem**.

According to The Information, OpenAI is planning to launch an AI model application store similar to Apple's "App Store", allowing developers to put their products based on OpenAI technology, such as chatbots or customized for various vertical fields model etc.

Many of ChatGPT's corporate customers usually customize AI models for their specific purposes, such as identifying financial fraud, or answering market-specific questions based on internal documents. As soon as there are more customized models, OpenAI has the idea of building a model store, so that developers or startups of such models can provide them to other companies through the OpenAI platform.

It's unclear at this time if OpenAI will charge a commission for transactions in the store, or otherwise generate revenue from the store.

It is reported that among OpenAI's customers, the enterprise AI platform company Aquant and the online education service provider Khan Academy are interested in joining this store and providing their AI models developed based on ChatGPT.

The report pointed out that OpenAI CEO Sam Altman disclosed the potential plan to some developers during a meeting last month. However, OpenAI responded that the company has not yet taken "aggressive action" to develop an app store.

In fact, outside of OpenAI, many companies have begun to build AI application stores to build their own AI ecosystem.

For example, Microsoft sells GPT enterprise services to customers through Azure cloud services, and Salesforce has AppExchange to help other companies sell applications.

It is worth mentioning that OpenAI has previously launched the ChatGPT Plugin, which can integrate ChatGPT into websites or applications, but the market response was mediocre. Sam Altman later admitted that this product did not find a fit with the market.

In addition, a few days ago, OpenAI also announced a major update to the API of the GPT model. The update includes: improving the long text processing capability of the model, realizing the conversion of natural language queries into API calls, and reducing API call costs.

On the other hand, many A-share companies revealed that they have connected/planned to connect to OpenAI’s GPT model API, including Tianyu Digital, Tom Cat, Montnets Technology, Zhidu, Xuanya International, Liaison Interactive, Kingsoft Office, Blue Cursor etc.

Ecological co-construction is an effective path for AI development

As Tang Daosheng, CEO of Tencent Cloud and Smart Industry Business Group, said, ecological co-construction is an effective path for AI development.

The ecological co-construction is inseparable from the prosperity of downstream applications. Everbright Securities pointed out that the key to winning the large-scale model competition lies in the control of downstream applications**. Huatai Securities pointed out that large model + privatization knowledge is expected to become an important landing direction of GPT + 2B applications, and 2B simple links/applications may be the earliest angle of entry.

Minsheng Securities believes that AI has entered an inflection point from the supply side to the application side, and at this stage it is expected to replicate the path from the supply side of TMT in 2010-2012 to the application side of TMT in 2013-2015. Different from the release of pure large-scale models in February and March, the new wave of releases is that the application products based on large-scale models have begun to be upgraded and launched on a large scale. In preparation for entering thousands of households, AI large-scale model products and applications are expected to usher in a concentration in June. release.

**As for how to find AI applications worthy of attention? **Guosen Securities listed several indicators:

1. Product landing speed: whether relevant AI applications have been launched in existing business scenarios, whether AI functions can meet the needs of existing scenarios, and drive payment rate and ARPU value to increase;

2. Segmentation scenarios have data advantages: For the same scenario, whether the data of application manufacturers is scarce is very important;

3. Explosive breakthroughs in user indicators & financial indicators: After accessing the large model, can you observe explosive growth in the number of users or financial indicators;

4. Business model change: whether to fully open the revenue space by introducing AI functions.