Google has unveiled Gemini Pro, the latest addition to its AI model family, aimed at empowering developers and enterprises to harness its capabilities for various use cases.
This release comes shortly after Google introduced Gemini, its most substantial AI model, featuring three sizes: Ultra, Pro, and Nano.
The company has already integrated Gemini Nano into its Android system, starting with the Pixel 8 Pro, and a customized version of Gemini Pro in Bard. Today, Google is making Gemini Pro accessible for developers and enterprises, offering a glimpse into its remarkable potential.
Gemini Pro: A game-changing AI model
Gemini Pro stands out in the AI landscape, demonstrating its prowess by outperforming similarly-sized models on research benchmarks. In its initial release, Gemini Pro boasts a substantial 32K context window for text, promising even larger context windows in future iterations.
Google has made it clear that Gemini Pro is available for free, with certain limitations, and assures competitive pricing as it evolves.
This robust AI model comes packed with a wide range of features, including function calls, embeddings, semantic retrieval, custom knowledge grounding, and chat functionality.
Gemini Pro is language-friendly, supporting 38 languages across 180+ countries and territories worldwide. In its current form, Gemini Pro accepts text as input and generates text as output.
Additionally, Google has introduced a dedicated Gemini Pro Vision endpoint that accepts both text and imagery as input, expanding its usability for multimodal applications.
Developers can harness the power of Gemini Pro through software development kits (SDKs) available for various platforms, including Python, Android (Kotlin), Node.js, Swift, and JavaScript. These SDKs make it easier for developers to create applications that can run seamlessly across different environments.
Introducing Google AI Studio: A developer-friendly tool
To further facilitate the development process, Google has introduced Google AI Studio, a web-based developer tool designed to streamline the creation of prompts and obtain API keys for app development.
Developers can access Google AI Studio with their Google accounts and enjoy an impressive free quota, allowing for 60 requests per minute – a significant increase compared to other free offerings.
Google AI Studio’s integration into the development workflow ensures a smoother transition from ideation to implementation.
However, it’s essential to note that when utilizing the free quota, developers should be aware that their API usage and Google AI Studio inputs and outputs may be accessible to trained reviewers. This data is meticulously de-identified to protect user privacy.
Seamless transition to Vertex AI on Google Cloud
For those seeking a fully managed AI platform, Google offers a straightforward transition path from Google AI Studio to Vertex AI. This transition allows users to customize Gemini Pro while maintaining full control over their data.
Vertex AI offers additional Google Cloud features that cater to enterprise security, safety, privacy, data governance, and compliance.
With Vertex AI, users can fine-tune and distill Gemini with their company data, integrate real-time information, and extend the model’s capabilities for practical applications.
This includes the creation of Gemini-powered search and conversational agents in a low-code or no-code environment, supporting retrieval-augmented generation (RAG), blended search, embeddings, and conversation playbooks.
One notable aspect of Google’s approach is its commitment to data privacy. Google affirms that its models are never trained on inputs or outputs from Google Cloud customers, ensuring that your data and intellectual property remain secure.
Gemini Pro pricing and future outlook
Currently, developers can access Gemini Pro and Gemini Pro Vision for free through Google AI Studio, with a generous rate limit of up to 60 requests per minute, making it suitable for most app development needs.
For Vertex AI developers, the same models are available with the same rate limits at no cost until the anticipated general availability early next year.
Following this period, Google will introduce a pricing structure based on usage, charging per 1,000 characters or images, applicable to both Google AI Studio and Vertex AI.
Looking ahead, Google is committed to refining Gemini Pro based on user feedback. The company plans to launch Gemini Ultra, its most advanced model, for highly complex tasks after conducting further fine-tuning, safety testing, and gathering valuable insights from partners.
Google also intends to expand Gemini’s integration into its developer platforms, including Chrome and Firebase, paving the way for innovative applications across various domains.
