Gemini: Google’s Powerful Generative AI Model
Google has recently announced the availability of Gemini, its most powerful and capable generative AI model, to enterprises for their app development needs. This move is aimed at enhancing the capabilities and accessibility of AI technologies in various industries.
Gemini comes in three sizes: Ultra, Pro, and Nano, with the Pro version being made accessible via API. Although currently available for free, there are certain usage limitations, as mentioned in a blog post by the company.
- “We’ve also made a dedicated Gemini Pro Vision multimodal endpoint available today that accepts text and imagery as input, with text output,” mentions Google.
- Sundar Pichai, CEO of Google, highlights that with the Gemini API, developers gain access to a wide range of features including function calling, embeddings, semantic retrieval, custom knowledge grounding, and chat functionality.
- Gemini Pro supports 38 languages and is accessible across 180+ countries.
We’ve also made a dedicated Gemini Pro Vision multimodal endpoint available today that accepts text and imagery as input, with text output.
Gemini Pro and Gemini Pro Vision in Google’s AI Ecosystem
Google AI Studio provides developers with a web-based platform to develop prompts and access an API key for app development using Gemini Pro. The Gemini API in the Google AI Studio is enhanced with a 32K context window for text generation, with plans for expansion in the future.
Gemini Pro is not limited to just the AI Studio, as it will be available on Vertex AI, Google Cloud’s comprehensive AI platform. Developers have the option to transition to a fully managed environment with Vertex AI, which includes tooling, fully-managed infrastructure, and built-in privacy and safety features for AI development.
Google plans to learn from developer feedback and fine-tune Gemini Pro, leading to the launch of the more advanced Gemini Ultra next year, designed for complex tasks.
We’ve also made a dedicated Gemini Pro Vision multimodal endpoint available today that accepts text and imagery as input, with text output.
As of now, Gemini Pro and Gemini Pro Vision are accessible for free with a rate limit of up to 60 requests per minute. This applies not only to developers using the models on Vertex AI but also to those using the Gemini API in the Google AI Studio. However, once the offering is generally available next year, Google plans to charge per 1,000 characters or per image for usage on both platforms.
The input price of Gemini Pro is set at $0.00025 per 1K characters and $0.0025 per image, while the output price remains constant at $0.0005 per 1K characters. It’s worth noting that Google’s pricing model charges “per character,” unlike some competitors who charge “per token,” which represents entire words.
Updates to Vertex AI and Introduction of Imagen 2
Vertex AI, Google Cloud’s end-to-end AI platform, has been updated with Imagen 2, the latest text-to-image diffusion technology developed by Google. Imagen 2 brings several new features, including the ability to generate various creative and realistic logos, emblems, and lettermarks. It also improves results in areas where text-to-image tools often face challenges, such as rendering text in multiple languages.
Additionally, Google is offering MedLM, a family of foundation models fine-tuned for the healthcare industry, to US-based organizations through Vertex AI. MedLM builds upon the earlier Med-PaLM 2 model released this year and is expected to receive a Gemini-based upgrade in the near future.
“We’ve also made a dedicated Gemini Pro Vision multimodal endpoint available today that accepts text and imagery as input, with text output,” mentions Google.
With these advancements, Google is pushing the boundaries of AI technology and making it more accessible to developers and enterprises. The availability of Gemini Pro and the updates to Vertex AI further empower users to harness the potential of AI in their app development and industry-specific projects.
