Microsoft makes Azure OpenAI Service generally available

Throughout 2022, large language models and AI models like Dall-E and ChatGPT seized headlines. Now, Microsoft is working to cater to market demand in this segment by announcing the general availability of Azure OpenAI Service.

Going forward, customers will be able to leverage Azure infrastructure to run advanced models like Codex, Dall-E, and GPT-3.5, along with the optimized service ChatGPT.

Azure OpenAI Service first debuted in preview in November 2021, with a broader limited access preview in mid-2022. So far, customers have used these capabilities to generate content, summarize customer service tickets, and optimize employee working time.

Microsoft began its current wave of AI language models with breakthroughs in speech recognition and machine translation between 2016 and 2018. It partnered with OpenAI in July 2019, investing $1 billion and exclusively licensing GPT-3 models and building AI supercomputing capabilities throughout 2020.  

According to Microsoft CVP for AI Platform, Eric Boyd, Azure is the only cloud that supports massive scale-up AI supercomputers. Microsoft tries to deliver a high level of performance with GPU and networking capabilities, and its infrastructure is used by several AI companies like Open AI, Hugging Face and Meta. OpenAI Service uses technology similar to that used by Microsoft in other products like GitHub Copilot, Power BI, and Microsoft Designer.

Boyd highlighted other examples of companies using OpenAI Service. Moveworks, a startup, uses the service to spot internal knowledge bases and to draft internal articles for customers, whereas Al Jazeera Digital is applying the service for summarization, translation, and AI tagging. KPMG, in turn, is using the service to improve tax efficiency and meet ESG goals.

Customers can apply for access to OpenAI Service and once approved, log in through Azure portal to create a resource.