Google releases Gemini Pro, new AI tools for developers and enterprises

Gemini Pro will be made available to developers through Google AI Studio and Vertex AI

Google is releasing a version of its latest Gemini artificial intelligence (AI) model - Gemini Pro - to developers and enterprises on December 13, along with a range of new AI tools, models, and infrastructure. This comes a week after the tech giant unveiled Gemini, its flagship multimodal AI model.

This launch comes as Google looks to attract more developers to boost the growth of its cloud offerings amid intense rivalry with Microsoft.

Gemini Pro will be made available to developers through Gemini API in the company's free web-based developer tool Google AI studio and to enterprises through Google Cloud's fully managed AI platform Vertex AI.

Developers will also have an option to transition their AI Studio code to Vertex AI for additional customization and other Google Cloud features in the future.

Google said the AI model supports 38 languages across more than 180 countries and territories worldwide. It currently accepts text as input and generates text as output. The company said it has also made a dedicated Gemini Pro Vision endpoint that accepts text and image input, and text as output for multimodal use cases.

During the Gemini launch, Google said that Gemini Pro outperformed GPT-3.5 in six of eight benchmarks including in MMLU (massive multitask language understanding) benchmark, which uses a combination of 57 subjects such as math, physics, history, law, medicine and ethics for testing both world knowledge and problem-solving abilities and GSM8K (Grade School Math 8K), which measures grade school math reasoning.

Developers will have free access to Gemini Pro and Gemini Pro Vision through Google AI Studio, with up to 60 requests per minute. Vertex AI customers will also have access to the models with the same rate limits, at no cost until general availability early next year, after which there will be a charge per 1,000 characters or per image across Google AI Studio and Vertex AI.

"The combination of the efficiency in our AI infrastructure and in Gemini itself along with many other software optimizations means that Gemini runs in very low latency" said Google Cloud CEO Thomas Kurian in a media briefing.

"It also allows us to offer attractive pricing to developers. Our new Gemini model is priced four times less on input characters, and two times less on output characters, compared to where we were in June this year" Kurian said.

He mentioned that the number of active generative AI projects on Vertex AI grew by more than 7x between Q2 2023 and Q3 2023.

Read: Google’s ‘multimodal’ AI model Gemini: What it means for enterprises?

New AI models

In addition to Gemini Pro, Google is introducing an upgraded version of its image AI model - Imagen 2 - which the company mentions is the most advanced text-to-image diffusion technology from Google DeepMind to date.

Imagen 2 delivers significantly improved image quality and text rendering in multiple languages along with other features such as the ability to generate a range of creative and realistic logos — including emblems, lettermarks and abstract logos — for business, brands and products.

Read: Google unveils Gemini, its largest AI model, to take on OpenAI

The AI model is also adding support for six additional languages including Hindi, Japanese, Korean, Portuguese, Chinese and Spanish in preview with plans to expand to more languages in early 2024.

The tech giant also introduced MedLM, a family of foundation AI models fine tuned for the healthcare industry, that will be available to Google Cloud customers in the United States through Vertex AI. MedLM builds on the Med-PaLM 2 foundation model introduced earlier this year and the firm said it will soon add Gemini-based models into the MedLM suite.

Besides this, Google also announced the general availability of two new capabilities on Duet AI in Google Cloud, its AI-powered collaboration solution across Google Cloud and IDEs (Integrated Development Environments).

Duet Al for Developers will help users code faster with Al code completion, code generation, and chat in multiple IDEs, thereby streamlining repetitive developer tasks and processes with shortcuts for common tasks. This availability also marks the company's entry into the AI-powered developer productivity tool market.

The company said that more than 25 code-assist partners such as MongoDB, Confluent, and knowledge-base partners like Atlassian, Datadog, JetBrains will contribute datasets specific to their platforms, so that developers can receive Al assistance based on partners' coding and data models, product documentation, best practices, and other useful enterprise resources.

Read: Gemini vs GPT4: Is Google’s new AI model as good as it claims to be?

Meanwhile, Duet AI in Security Operations can enable defenders to more effectively protect their organisations from cyberattacks

Over the next few weeks, Google said it also plans to integrate Gemini across its Duet AI portfolio.

"Duet AI for Developers and Duet AI in Security Operations are the first two Duet AI in Google Cloud offerings to go general availability, with more planned for early next year, including Duet AI in BigQuery, Looker, our database products, Apigee, and Colab Enterprise" Brad Calder, vice president and general manager, Google Cloud Platform said in a blogpost.