Azure AI Studio has officially launched, offering companies the ability to create and enhance AI models and applications. The tool was announced by Microsoft CEO Satya Nadella during the Build 2024 event, where developers were updated on the latest developments from Microsoft. With Azure AI Studio, users can build, train, and refine AI models for various applications.
During the preview phase, Azure AI Studio underwent improvements to address AI vulnerabilities such as Prompt Shields, inaccuracies, and jailbreaking. The tool provides an end-to-end solution for developing and securing AI applications. Additionally, it now includes over 1,600 AI models, including the recently launched GPT-4o model by OpenAI. Other models in the catalog come from Meta, Mistral, and Microsoft.
One of the new models featured in Azure AI Studio is Phi-3-vision, which combines language and image processing capabilities. This multimodal model is part of the Phi family and contains 4.2 billion parameters. Microsoft claims that Phi-3-vision has fewer parameters compared to other large language models (LLMs) developed by leading companies.
Overall, Azure AI Studio provides companies with a comprehensive tool for creating advanced AI models and applications. The launch of new models, including GPT-4o and Phi-3-vision, expands the capabilities of the platform and offers users the opportunity to explore innovative AI solutions. The tool’s focus on addressing AI vulnerabilities and providing end-to-end development support makes it a valuable resource for companies looking to leverage AI technology in their products and services.
Article Source
https://www.techzine.eu/news/applications/120212/azure-ai-studio-made-generally-available-with-gpt-4o-and-phi-3-vision/