On May 8, Google (GOOG) DeepMind AI research department officially launched Gemini 2.5 Pro “I/O” special edition, responding to the developer community’s strong demand for actual programming and interface design performance.
Leading the test, the strongest programming model is coming
It is said that this special edition is an upgraded version of the popular model Gemini 2.5 Pro released in March. Google Gemini API senior product manager Logan Kilpatrick confirmed in a technical blog that this upgrade focuses on optimizing function call functions, reducing error rates and improving trigger reliability.
Gemini 2.5 Pro “I/O” special edition also has ultra-high performance in video understanding. Google said in its blog that this version scored 84.8% in the VideoMME benchmark test, ranking first. According to Google’s official blog, the model also drives core functions such as the Canvas canvas of the Gemini mobile application.
In addition, Google surpassed Claude in key programming benchmarks for the first time, achieving an all-round lead. The new model codenamed “Gemini-2.5-pro-preview-05-06” has replaced version 03-25 and is currently available through platforms such as the Gemini API. Ordinary users can experience it through the Gemini chatbot application.
Tencent upgrades the dual model of Hunyuan + DeepSeek
At the same time, on May 7, Tencent (TCEHY) AI assistant Yuanbao announced that its Wenshengtu function has ushered in a major upgrade. The biggest highlight of this upgrade is the introduction of Tencent Hunyuan’s latest Wenshengtu model, and supports Hunyuan and DeepSeek models for image generation. Users only need to enter a sentence command to generate high-quality and imaginative pictures.
Tencent said that the upgraded Wenshengtu function has a wide range of application scenarios. In the field of artistic creation, it can inspire artists and help them quickly generate images with creativity and artistic style. In addition, this function can also meet the personalized needs of users and generate images that meet specific scenarios.
Nowadays, generative artificial intelligence represented by ChatGPT and DeepSeek has set off a new round of industrial revolution, which in turn has triggered the “intelligent emergence” of large computing power, large models, and big data, ushering in a window period for the development of AI large models.
It is worth mentioning that relevant policy measures will be further strengthened. The Ministry of Industry and Information Technology recently stated that my country has formed a complete artificial intelligence industry system covering the basic layer, framework layer, model layer, and application layer. The next step will be to strengthen the research and development layout of general large models and industry large models, and accelerate the construction of high-quality data sets in the industrial field.
The next step will be to fully promote artificial intelligence to empower new industrialization, focus on supply and application, and accelerate the implementation of artificial intelligence large models in electronics, raw materials, consumer goods and other industries, and be applied in R&D design, pilot verification, production and manufacturing, and operation management. At the same time, accelerate the cultivation of a group of industry leading enterprises and specialized and new small and medium-sized enterprises, and build an artificial intelligence open source community.
WiMi accelerates the empowerment of AI model technology
Undoubtedly, since ChatGPT detonated the generative AI competition, various models have become an important innovation direction for artificial intelligence. According to the data, Wimi Hologram Cloud Inc (WIMI), as a company that has invested in AI technology research and development for a long time, has deeply optimized open source model frameworks such as DeepSeek under the optimization of the team’s new technology, enabling it to achieve significant performance improvements in a variety of application environments.
The long-term and continuous investment in AI technology has made WiMi always at the forefront of the field of artificial intelligence. At present, its technology has been fully open source, and the model code, computing resources and technical interfaces have been opened, which have been successfully applied to the training and reasoning of various large models and other projects, supporting third-party developers to call computing resources for model training, lowering the technical threshold, and allowing new technologies and new applications to continue to emerge.
In addition, WiMi provides AI model-driven efficiency optimization solutions for scenarios such as e-commerce live broadcasting and digital marketing, such as virtual anchors replacing real-person operations to reduce corporate labor costs. At present, relying on technological advantages to participate in market competition has not only strengthened WiMi’s leading position in AI vision, digital humans and other fields, but also further demonstrated its own intention to provide the industry with full-chain support from large model training to application landing.
Summary
A new round of scientific and technological revolution and industrial transformation is reshaping the global innovation landscape, especially the emergence of artificial intelligence big models, which marks that artificial intelligence has begun to become an epoch-making protagonist on the human science and technology stage. Major cutting-edge technologies and disruptive technologies are rapidly breaking through globally. Artificial intelligence big models are targeting industry applications and deeply integrating with electronics, raw materials, consumer goods and other industrial fields, promoting cost reduction and efficiency improvement in segmented business scenarios. The role of profound adjustment of innovation methods in the future will be more prominent.
