Google released Gemini for consumers last week, and this week it brings Gemini Pro to developers and businesses, a powerful AI service provided through AI Studio and Vertex AI.

The first version of Gemini Pro supports text input/output with a 32K context window, which Google says will be increased in the future. Additionally, there is an endpoint for Gemini Pro Vision, which can handle both text and image input and output text. Google also highlights some features of Gemini Pro:

• It offers various functionalities, including function calls, embedding, semantic retrieval, custom knowledge bases, and chat capabilities.
• Gemini Pro provides an SDK to help you build applications that can run anywhere. It supports Python, Android (Kotlin), Node.js, Swift, and JavaScript.
• It supports 38 languages in over 180 countries and regions worldwide.

Gemini Pro and Gemini Pro Vision are now available for free through Google AI Studio (formerly known as MakerSuite). This web-based tool is designed for individual developers and teams, allowing you to quickly develop prompts and obtain an API key for application development. You can log in with your Google account and easily import files from Google Drive.

The "Get Code" button allows you to transfer your work to your IDE, supporting cURL, JavaScript, Python, Android (Kotlin), and Swift.

Today's release offers a "free quota" that allows 60 requests per minute, which Google claims is 20 times more than other free products. Google will use these free inputs/outputs to improve the model, and the data will be anonymized with your Google account and API key. It may be accessed by trained reviewers.

Vertex AI from Google Cloud is a more advanced product that provides greater control and integration. As a fully managed AI platform, you can use your own data to build search/conversational agents based on Gemini. Additionally, Google will not train models on inputs or outputs from Google Cloud customers.

Access to Gemini in Vertex AI will remain free until the early next year. At that time, Google AI Studio and Vertex AI will be priced at $0.00025 per 1K characters of input, $0.00025 per image input, and $0.0005 per 1K characters of output. Google attributes these prices to its efficient TPUs.

Looking ahead, Google plans to bring Gemini to its other development tools, including Android Studio, Colab, Firebase, and Flutter. On the consumer side, the company revealed that Gemini will debut on Duet AI for Workspace in early 2024.

Google also announced Imagen 2 today, which features "significantly improved image quality" and the following capabilities:

• Generating high-quality, realistic, high-resolution, and aesthetically pleasing images based on natural language prompts.
• Multi-language text rendering to create images with accurate text overlays.
• Logo generation to create company or product logos and overlay them on images.
• Visual question answering to generate captions from images and provide informative text answers to questions about image details.