According to news on December 14, Google announced a series of upgraded artificial intelligence (AI) features on Wednesday, U.S. time, aimed at providing better services to its cloud computing customers. The tech giant is trying to catch up with rivals such as Microsoft and OpenAI, which are actively capitalizing on the artificial intelligence boom.
According to reports, Google has released GeminiPro for enterprises, allowing developers to build applications using Google's latest artificial intelligence models. Gemini is a large-scale artificial intelligence system trained with large amounts of data to generate new content based on user requests.
Last week, Google launched Gemini, which it claimed was the first large language model to surpass OpenAIGPT-4 in many areas, and now it has launched GeminiPro for enterprises.
Google Cloud customers can use GeminiPro to create applications such as artificial intelligence chatbots, easy-to-query inventory databases, and marketing presentations. The company also emphasized that GeminiPro will initially be available to cloud customers for free, with some limitations. However, Google said it ultimately plans to ensure that its cloud AI products are "competitively priced."
Parameters released by Google show that GeminiPro's text-based function has an input cost four times lower and an output cost two times lower than its previous generation artificial intelligence model PaLM2 released in June.
Google Cloud CEO Thomas Kurian said the AI model was built to "generally and seamlessly understand, manipulate and combine different types of information, including text, code, audio, images and video, just as humans see, hear, read, listen and talk about different types of information simultaneously."
Although Google is a pioneer in the field of generative artificial intelligence, the popularity of its products lags far behind that of its rivals. Last week, Google tried to counter that narrative with the release of Gemini. Gemini has three models, namely GeminiUltra, GeminiPro and GeminiNano. Google has begun rolling out the smallest Nano version that can run directly on devices such as Google's flagship smartphone Pixel 8 Pro.
At the same time, the company also released a customized version of GeminiPro, Google's artificial intelligence chatbot designed to compete with OpenAI's popular ChatGPT. By releasing GeminiPro to app developers and enterprises, Google hopes to send a message that it is no longer behind OpenAI and its latest artificial intelligence system, GPT-4, which is accessible through partner Microsoft's Azure cloud computing platform.
Google said GeminiPro supports 38 languages in 180 countries and regions around the world, currently accepts text as input and can generate text output. Google also released a dedicated GeminiProVision platform that can handle text and image-based prompts from users.
The company also said GeminiUltra is the company's largest and most powerful model for "highly complex" tasks and will be available in early trials to select cloud customers and partners before being released to the public next year.
Google announced that Gemini Pro will be integrated into two key cloud products, Google AI Studio and Vertex AI. Among them, Google AI Studio is a free web-based developer tool that Google calls “the fastest way to build using Gemini.” This tool allows customers to develop applications using GeminiAPI. At the same time, VertexAI provides more customized services for developers and cloud customers. Enterprises will be able to customize Gemini with their own data and build applications such as Gemini-based search tools and chatbots.
Kurian said GeminiPro's pricing is becoming "more attractive." The company said developers will be able to use Gemini Pro and Gemini Pro Vision for free through Google AI Studio, which is suitable for most application development needs. The more flexible VertexAI will be free until early next year.
Google also announced an upgraded version of its previously released artificial intelligence model, Imagen2, Google's text-to-image technology, which will improve photo realism, text rendering and logo generation capabilities. Google also launched MedLM, a series of models fine-tuned for the healthcare industry, building on the company's work on Med-PaLM2. Med-PaLM2 is Google’s artificial intelligence model, which has been trained with professional medical knowledge.
Google also announced a global partnership with MistralAI, a Paris-based artificial intelligence startup focused on open source software. MistralAI will distribute some of its artificial intelligence products, including optimized proprietary language models, on Google Cloud's infrastructure, the companies said in a blog post.
While the agreement is not exclusive, MistralAI CEO and co-founder Arthur Mensch praised Google Cloud's flexible tools and ability to support the company's products. "Google Cloud's open source support and principles for responsible development of AI technology, broad and reliable infrastructure capabilities, and commitments around privacy and security are very consistent with our mission to develop open and available models," he said.
With the recent launch, Google is positioning Gemini as "our next step into artificial intelligence" and touting it as having "state-of-the-art capabilities." This could be an attempt by Google to divert attention away from OpenAIChatGPT, or to go head-to-head with them. If this happens, they will be closer to what they want to achieve. But for now, these products need to perform well and resonate with consumers, including cloud customers.