Microsoft: We’re bringing ChatGPT to the Azure cloud-computing service

Microsoft has announced general availability of its Azure OpenAI Service, an offering related to its $1 billion investment in OpenAI, the maker of ChatGPT. 

That means more businesses can apply to use OpenAI’s Azure-hosted and trained large language models (LLMs), such as: GPT-3.5, the LLMs behind ChatGPT; DALL•E 2, the AI that generates images from casual descriptions; and Codex, the GPT-3-based foundation of GitHub’s Copilot AI paired programming service. 

Microsoft invested $1 billion in OpenAI in 2019, and in return OpenAI has built its AI models on Microsoft’s Azure AI supercomputing technologies as it pursues General Artificial Intelligence. OpenAI has been scouting for investors at a valuation of $29 billion and Microsoft was reportedly in talks to take a $10 billion stake in the firm at a valuation of $29 billion. Shortly after opening ChatGPT in November, OpenAI’s chief said ChatGPT responses cost in the single-digit cents. The chatbot gained one million users within a week of its launch.    

Microsoft corporate vice president Eric Boyd says in a blogpost that Microsoft Azure customers will be able to access ChatGPT through Azure OpenAI Service soon. ChatGPT has been trained and runs inference models on Azure AI infrastructure, he notes. 

Also: You can use an AI Time Machine to see what you’d look like in different eras throughout history

In related news, OpenAI announced that it is adding ChatGPT to its roster of APIs, so that third-party developers can interface their applications with it. Those interested will have to join a waitlist, however.

Microsoft debuted the Azure OpenAI Service in November 2021 as a private preview for select customers. It points to several customers already using Azure OpenAI Service, including KPMG, Moveworks, and Al Jazeera. 

Microsoft notes that Power BI is already using GPT-3 to generate formulae and expressions. Microsoft is also using DALL•E 2 to power its Microsoft Designer AI-generated images.     

Microsoft already runs these services and GitHub’s Copilot on the Azure OpenAI Service.

Microsoft has created tiers of GPT-3, DALL•E 2, and Codex, with different capabilities, which are called Davinci, Curie, Babbage, and Ada. Davinci is the most capable, but others are faster.