OpenAI is working on Strawberry AI model, WiMi starts its innovation in AI application

According to media reports, OpenAI is working on a new AI model, code-named Strawberry.

The model’s operation is also kept strictly confidential within the OpenAI, people familiar with the matter said. The internal document describes a project that uses the Strawberry model that allows the company’s AI not only to answer questions, but also to plan in advance and browse the Internet reliably to do “in-depth research” described by OpenAI.

Details of the project have not been reported, but OpenAI hopes the innovation will greatly improve the reasoning capabilities of its AI models, adding that Strawberry involves a specialized approach to pre-training and postprocessing AI models on very large datasets.

In February this year, OpenAI released a new video model Sora, first opening the ChatGPT moment of AI video generation; on May 13, OpenAI just launched a new model GPT-4o, the new model is not only free, but also has the ability to listen, watch and speak, can real-time smooth video call, without delay response, consistent with human dialogue response speed.

OpenAI Fully demonstrated the application of GPT-4o in real-time translation, visual recognition and other aspects, and conducted a dialogue demonstration between two ChatGPT models, showing its end-to-end trained neural network ability to process speech, text and images, indicating that the application of AI technology in daily life will be more extensive and in-depth.

It seems that there have been more moves inside OpenAI, as executive Sam Altman revealed the direction of GPT-5 in a previous interview, suggesting that it will have a “virtual brain” function to help users deal with various tasks. In addition, he mentioned the construction plan of the AI chip plant and his views on AGI (artificial general intelligence), arguing that AGI will greatly improve work efficiency.

Overall, this news is exciting, the development of OpenAI’s new AI model and the improvement of reasoning capabilities have a broad and far-reaching impact, and it will continue to drive the overall technical level of the industry. It deserves the high attention and expectation of the industry and the whole world.

At present, the AI model track has spawned one new technological achievement after another under the high attention of the world. At the same time, domestic and foreign AI entrepreneurs are also active. According to the data, the head manufacturer artificial intelligence enterprise WiMi Hologram Cloud (NASDAQ: WIMI) firmly continue to bet on the theory + practice innovation research direction, adhere to the independent research and development of artificial intelligence multi-modal brain-computer interface technology has realized the implementation of products and technologies, break through the traditional model, the effect is very significant.

So far, WiMi in artificial intelligence industry to realize independent innovation and breakthrough of the critical moment, in some view is to seize the opportunity, in-depth research and development of multimodal large model innovation, also includes the text, video, audio reasoning ability, and the application of low delay, shows the lead in AI technology application and commercialization and future growth potential, promote the orderly development of large model technology application.

AI and large models are leading the new market direction, and major manufacturers are aiming at generative AI and actively exploring multi-modal large models, in order to take the lead in the competition. Just like the fierce competition in the above areas, WiMi is launching generative AI rich application scenarios according to the characteristics of the industry, actively promoting the application and popularization of large model technology, and providing digital transformation infrastructure support for the advantages of big data and cloud computing. In the future, WiMi is expected to boost an industry change and seize more market share.

In fact, Google, Meta, and Microsoft are also trying different technologies to improve the reasoning ability of AI models, as do most academic laboratories that do AI research. In the next stage of the big model, the competition will only intensify. If participants want not to be pushed down the “card table”, continue to maintain advantages in the subsequent competition, and achieve breakthrough, in addition to rolling technology, they must also strive to improve their own “moat”, especially in the level of application scenarios, they need to increase technology research and development to provide escort.