Google finally takes on GPT-4: The all-rounder Gemini is finally on the scene!

On the morning of December 6, Silicon Valley time, Google CEO Splinter announced that after months of dedicated research and development, Google’s new multimodal large model Gemini is officially online. This model not only shows strong ability in text, image, video, audio and code modalities, but also surpasses GPT-4 in a number of performance, and is regarded as the most promising model to surpass GPT-4.

Gemini is a native multimodal large model, Google in May this year’s I/O conference announced the beginning of the development of the Gemini legend continues: the merger of the Google Brain and DeepMind department, hundreds of people to attack, almost exhausted Google’s internal computing resources …… so all these things, just to and OpenAI battle! Gemini is a new technology that has been in development for a long time.

But it was only when OpenAI’s GPT-4 went live after half a year and Silicon Valley blew up the circle that Gemini came out in the midst of a thousand calls. Now, it finally unveiled its mystery, showing its five major capabilities of text, image, video, audio and code, and at the same time launched three versions of large, medium and small, from the cloud to cell phones and tablets can run.

Amidst the concerns of Jim Fan, a senior scientist at NVIDIA, Gemini shows its amazing power. It can not only process text information, but also understand image information, and can even interact with simple games. This all shows that Gemini has strong natural language processing and multimodal processing capabilities.

Not only that, but Gemini has a number of cool use cases: the AI can react accurately to a video, the AI can play you-draw-me-guess …… All of these show the potential of Gemini as a true human assistant.

In this release, Gemini finally unveiled its mystery – showing off its five major capabilities of text, image, video, audio and code. Three versions, large, medium and small, were launched at the same time, running from the cloud to cell phones and tablets. It not only understands and replies to human text messages, but also handles multimedia information such as images and videos, and can even do simple code writing and debugging. These capabilities give people a glimpse of Gemini’s promising future in the multimodal field.

In addition to this, Gemini has a number of eye-opening features. For example, it understands image information based on the image. This means that it does not need to use OCR technology to “recognize” the image first, and then put it into the language model for semantic understanding. This is an important feature of Gemini: end-to-end understanding, where information is not lost in the “transcription” process.

In the demo, Gemini’s performance was also impressive. Whether it was a simple conversation with the presenter or performing some complex tasks such as generating code or providing suggestions for a party event, Gemini was able to excel. This showed the usefulness and potential of Gemini.

To show its all-around strength, Google also conducted many performance tests. The results show that Gemini outperforms current state-of-the-art models on both natural language processing and multimodal tasks. This shows that Gemini is a very powerful and well-rounded model.

With the release of Gemini, Google is also trying to apply AI technology to more areas. The AlphaCode 2 launched this time not only understands, interprets and generates high-quality code in programming languages such as Python, Java, C++ and Go, but also solves programming competition problems that are beyond the scope of programming and involve complex math and theoretical computer science. This shows that Google is constantly exploring the application scenarios of AI technology and trying to apply it to real life.

Google DeepMind CEO Demis Hassabis said, “This is the largest and most powerful big model we have to date. gemini can understand the world around us, just like we do.” This shows that Google has very high expectations and pursuits for the development and application of artificial intelligence technology.

Overall, the release of Gemini is undoubtedly a major breakthrough for Google in the field of artificial intelligence. It not only shows Google’s cutting-edge technological strength in the field of multimodality, but also indicates that Google has a very high pursuit for the exploration and application of AI technology. In the future, we look forward to seeing the emergence and application of more advanced models like Gemini, which will bring more convenience and surprise to human beings.