Microsoft Azure AI Studio adds Llama 2, GPT-4 Turbo V as services

Microsoft, one of many largest winners of the generative AI growth thus far because of its early backing of OpenAI and integration of the latter startup’s tech into Bing, Azure, and numerous different companies, has clearly been attempting to keep away from placing all of its AI eggs in a single basket.

Right now, the company announced that it’s making Meta Platforms’ rival open-sourced AI model Llama 2 obtainable in Azure AI Studio, its cloud platform AI clearinghouse, within the type of a “model-as-a-service.”

Understanding AI models-as-a-service

Just like the now outdated acronym of SaaS (software-as-a-service), MaaS provides prospects the power to make use of AI fashions akin to Llama 2 on-demand over the online when wanted with minimal arrange, with out having to go to all the difficulty of putting in it on their very own cloud server house, personal cloud, hybrid cloud, or different location. Even with Microsoft consultants serving to and educated IT personnel, this generally is a non-trivial job for enterprises.

As John Montgomery, Microsoft Company Vice President of Program Administration for its AI Platform efforts defined in a blog post:

“In Azure AI, you could have been in a position to deploy fashions onto your individual infrastructure for a very long time – merely go into the mannequin catalog, choose the mannequin to deploy and a VM to deploy it on and also you’re off to the races. However not each buyer desires to consider working infrastructure, which is why at Ignite we launched Fashions as a Service, which operates fashions as API endpoints that you just name, a lot the best way you may name the Azure OpenAI Service.

Right now, we’re making Meta’s Llama 2 obtainable in Fashions as a Service by way of Azure AI in public preview, enabling Llama-2-7b (Textual content Era), Llama-2-7b-Chat (Chat Completion), Llama-2-13b (Textual content Era), Llama-2-13b-Chat (Chat Completion), Llama-2-70b (Textual content Era), and Llama-2-70b-Chat (Chat Completion).”

Hedging AI bets whereas giving prospects extra decisions

Providing this big selection of various open supply Llama fashions looks like a sensible transfer on Microsoft’s half, increasing the alternatives of AI for Azure cloud storage and repair prospects and giving them a far lower-cost possibility (the fashions themselves are free, although numerous implementations might not be) to Microsoft’s buddy OpenAI’s GPT-3.5 and 4 fashions.

In any case, Llama 2 arguably won the year with regards to generative AI since it’s the popular open supply possibility for a lot of customers and enterprises.

It follows studies of Microsoft CEO Satya Nadella, who was mentioned to be shocked and furious in regards to the prior OpenAI board’s abrupt resolution to terminate CEO Sam Altman again in November earlier than finally re-hiring him, stating that he wants to spread Microsoft’s AI bets more widely.

OpenAI nonetheless nicely represented

However that doesn’t imply that OpenAI is being overlooked within the chilly with right this moment’s bulletins. Quite the opposite: Montgomery additionally shared the information that OpenAI’s newest publicly accessible AI mannequin, GPT-4 Turbo with Imaginative and prescient — which provides the AI the capabilities of analyzing and describing pictures and visible materials — can also be now obtainable to prospects of Azure within the Azure AI Studio.

Already, prospects akin to Instacart and WPP are utilizing this mannequin, in response to Montgomery.

And, the Azure AI Studio consists of instruments for fine-tuning all of the fashions supplied.

At this level, because the AI cloud wars proceed to rage on, the one actual query is: what number of fashions will Microsoft be capable to add to Azure AI Studio? Is Mistral or Deci subsequent?

The information right this moment additionally follows Microsoft’s release yesterday of Phi-2, it’s personal small language mannequin (SML), although that is just for analysis functions and may’t be used for industrial instances. Nonetheless, Llama-2 and GPT-4 Turbo with Imaginative and prescient can.