Azure AI Foundry already has greater than 1,900 AI fashions, together with the most recent from OpenAI, Meta, and others. In the present day, Microsoft introduced that xAI’s flagship Grok 3 and Grok 3 Mini fashions are actually accessible on Azure AI Foundry.

The Grok 3 fashions are supplied as a direct, first-party providing, so they’re hosted and billed instantly by Microsoft. Prospects utilizing the Grok 3 fashions on Azure AI Foundry could have all of the service-level agreements (SLAs).

Enterprises now depend on totally different fashions relying on their use instances. Nonetheless, deciding on fashions, evaluating them for numerous use instances, and deploying them is a tough activity for any enterprise. To resolve this downside, Microsoft is introducing a brand new functionality in Azure AI Foundry that allows straightforward mannequin choice with automated mannequin routing, real-time efficiency monitoring, and extra.

The brand new Mannequin Router preview function in Azure AI Foundry will routinely choose the most effective OpenAI mannequin for prompts. Microsoft claims that this new function improves response high quality and reduces the general price. Because of automated analysis, A/B experimentation, and tracing in Foundry Observability, enterprises can roll again to confirmed fashions if new ones underperform.

To assist enterprises rapidly design, customise, and deploy AI options, Azure AI Foundry will supply new AI templates meant for frequent, high-value use instances and technical patterns.

Lastly, Microsoft is bringing the ability of small fashions accessible in Azure AI Foundry to Home windows 11 PCs and macOS. The Azure AI Foundry Native preview will enable builders to simply run AI fashions, instruments, and brokers instantly on-device. Foundry Native might be a part of Home windows AI Foundry and can work on the most recent Home windows 11 Copilot+ PCs operating Intel, AMD, and Qualcomm chipsets. Foundry Native will allow AI functions that do not require the Web, enhance privateness, and scale back prices.