Google Vertex AI Model Garden
Google Cloud declared in July that Meta’s Llama 3.1 open models would be available in Vertex AI Model Garden. Since then, businesses and developers have expressed a great deal of enthusiasm for using the Llama models in their constructions. It is pleased to announce that the latest generation of multimodal models from Meta, Llama 3.2, is now accessible on Vertex AI Model Garden.
A new generation of lightweight vision models designed for edge devices, Llama 3.2 is optimized for use cases requiring more individualized and private AI experiences. Using the most recent models:
- Llama gets multimodal: You may now reason on high-resolution images, such as charts, graphs, and image captioning, with Llama 3.2’s new 11B and 90B vision LLMs. This opens up new opportunities for interactive instructional tools, image-based search and content creation, and more.
- Small Llama: The 1B and 3B lightweight versions of Llama 3.2 are made to integrate easily with mobile and edge devices. These models enable you to create individualized, private AI experiences with low latency and resource consumption. Imagine knowledge retrieval, local AI agents, and multilingual summarization all on-device, all while protecting user privacy.
The new Llama models put system-level safety and ethical innovation at the forefront while giving accessibility, efficiency, and privacy top priority.
- You can easily experiment with, modify, and implement models like as Llama 3.2 with Vertex AI’s unified AI platform. With Llama 3.2 joining Vertex AI, its carefully selected library of more than 160 enterprise-ready first-party, open-source, and third-party models in Model Garden is now even larger. This makes it possible for you to choose the ideal models for your requirements by utilizing its adaptable and open AI ecosystem.
- This is a glimpse of the new 90B model that you can simply access through its Model-as-a-Service (MaaS) offering. You can instantly access the model with MaaS, customize it with powerful development tools, and deploy it with pay-as-you-go billing and fully managed infrastructure. The 90B will be widely accessible in the upcoming weeks. In the upcoming weeks, the 11B vision model will also be accessible as MaaS.
Vertex AI Model Garden will allow users to self-servicely deploy all four of the Llama 3.2 models.
Clients use Llama for building on Google Cloud
Shopify is using Llama on Vertex AI to streamline its data gathering procedures and provide data-driven insights to help businesses globally:
Shopify now has more efficient and easier access to high-quality data creation with Llama 3.1 on Google Cloud Vertex AI.
TransCrypts uses Google Cloud’s Llama to make its AI-powered financial advisor, Castello, available to thousands of users.
For TransCrypts, using Llama 3.1 on Google Cloud has changed everything.
Through conversational AI, recommendations, and analysis, BMC has included Llama 3.1 into its flagship BMC Helix platform to expedite IT service and operations management.
Using Google Cloud’s Llama 3.2
Using Vertex AI and Llama 3.2, you can:
- Try it with assurance: Without worrying about difficult deployment procedures, explore Llama 3.2’s potential with straightforward API calls and its extensive generative AI evaluation service within Vertex AI’s user-friendly environment.
- Adjust Llama 3.2 to your precise requirements: Utilizing your own data, fine-tune the model to create custom solutions that are suited to your particular requirements.
- Ground your AI in truth: Use Vertex AI’s various grounding and RAG options to ensure that your AI outputs are dependable, pertinent, and trustworthy. You can utilize Llama for generation, Vertex AI Search for enterprise information retrieval, and more to integrate your models with enterprise systems.
- Create and manage intelligent agents: Utilizing Vertex AI’s extensive toolkit, which includes LangChain on Vertex AI, create and manage agents driven by Llama 3.2. With Genkit’s Vertex AI plugin, you can incorporate Llama 3.2 into your AI experiences.
- Deploy without overhead: With flexible auto-scaling, pay-as-you-go pricing, and top-notch infrastructure built for AI, Llama 3.2 applications can be deployed and scaled more easily.
- Work within the boundaries of your company: Deploy with confidence with Google Cloud’s integrated security, privacy, and compliance features as well as support for Meta’s Llama Guard for the models. Furthermore, business controls offer the proper access restrictions to guarantee that users may access only approved models. One example of this is the new organization policy for Vertex AI Model Garden.
Launch Llama 3.2 on Google Cloud to get started
The diversity of enterprise AI ecosystems increases with each new advancement in AI models. Its collaboration with Meta demonstrates how dedicated both companies are to delivering top-notch innovation backed by a transparent and easily accessible AI ecosystem.
