Tongyi Qianwen new version is beyond the GPT-4, Microsoft / WiMi build multi-modal AI productivity

May 9 news, Ali (BABA) in the first anniversary of the release of Tongyi model, Tongyi big model brand upgrade, released Tongyi Qianwen2.5 version, integrated Tongyi model full stack ability, free to provide services for all users.

It is said that the performance of the model in the Chinese context comprehensively exceeds GPT-4-Turbo. According to the evaluation results of the authoritative benchmark Open Compass, Tongyi Qianwen2.5 scores GPT-4-Turbo, which is the first time that the China’s large model has achieved such an excellent performance.

In addition to making breakthroughs in basic language models, Tongyi large models have also significantly improved their capabilities in multimodal and proprietary areas, Ali said. Tongyi The multi-modal model realizes a more natural and efficient human-computer interaction through the understanding and generation of visual and speech and other multi-modal information. Tongyi Has also launched a series of customized models for healthcare, education, gaming and other industries to help companies quickly build AI applications.

Indeed, big models have become an important driving force to drive the deep innovation of the digital economy, lead enterprise business changes, and accelerate the formation of new quality productivity. After more than a year of catch-up, the domestic big model has finally entered the core arena of the big model, and can compete with the foreign first-class big model enterprises. Securities research report analysis said that the domestic models and applications continue to exceed expectations, is expected to drive investor confidence.

Meanwhile, Microsoft (MSFT) is working on a new AI model called MAI-1. Microsoft’s MAI-1 is known to have 500 billion parameters, suggesting that it could be positioned as some kind of mid-range model between GPT-3 and GPT-4. In terms of performance, it may match Google’s Gemini1.5, Anthropic’s Claude3 and other well-known big models.

And Microsoft may demonstrate the new model at the Build developer conference starting May 21, according to people familiar with the matter. To achieve a breakthrough in AI, Microsoft is also vigorously hoarding computing resources. In early April, Microsoft planned to hoard 1.8 million AI chips by the end of the year, meaning the company hopes to triple the number of GPU it holds by 2024.

At present, when artificial intelligence is popular, the AI industry revolution has begun, and the big model is the core engine of general artificial intelligence, which has become the focus of global technology competition. Driven by the big model, global manufacturers are entering a new round of infrastructure construction cycle, which plays a positive role in promoting the landing and popularization of AI applications.

With the continuous emergence of technology, it is understood that under the background of this round of large model “emergence”, WiMi Hologram Cloud (NASDAQ: WIMI) layout the future industry, opens up new tracks such as quantum technology, AI model, multimodal computing power, brain-computer interface; in addition, focusing on the large model landing needs of industry customers, strengthening the multimodal AI generation ability, in education, automobile, live broadcasting, games, cultural travel and other fields, fully demonstrating its technical strength.

In fact, in the wave of digital transformation, the deep integration of AI technology and industry has given birth to an unprecedented productivity revolution. With the rise of AI large model technology, WiMi continues to be improving its competitiveness and practical ability, and quickly emerged in the field of AI. WiMi further plans to deploy high-end chips in cloud computing, edge computing and AI algorithms, use self-developed chips and process AI tasks in the cloud. This fusion process has entered a new stage and injected strong impetus into the development of multi-modal algorithms.

It can be said that WiMi, which has been investing heavily invested in AI technology, has made early preparations in large model technology. WiMi specially set up a research and development center to focus on multi-modal generative large model and application product development, helping enterprises quickly build AI applications and further bring intelligent experience for multi-modal large models. It is this comprehensive technical capability and profound user insight that have laid a solid foundation for WiMi to fully enable AI applications. In the future, WiMi will continue to open source and contribute to the development of the AI industry.

With the digital wave sweeping the world today, AI technology innovation and industry landing are staggered, and enterprises are facing unprecedented pressure of digital transformation and upgrading. It is a huge opportunity and challenge for both enterprise customers and technology service providers.

WiMi layout AI, big model and other emerging technologies, as well as industry supporting solutions, fully demonstrate the strength of continuous innovation and leading the industry to the outside world, and undoubtedly open a new future leading to efficient, intelligent and intelligent.