Please rotate your device

To continue exploring this website, please rotate your device to a vertical orientation

News

Baseten is now available on Microsoft Foundry

Enterprises on Azure can now run frontier open models powered by Baseten in Microsoft Foundry.

Published
Baseten is now available on Microsoft Foundry
TLDR

Baseten is now available in public preview on Microsoft Foundry, bringing frontier open models to Azure customers as a first-party Microsoft service. Teams can test, deploy, and scale these models, powered by Baseten’s optimized inference infrastructure, directly from the Foundry model catalog. For enterprises, that means more intelligence per dollar with the governance, reliability, and unified billing of Azure services.

More model choice for the next wave of AI innovation

Our partnership with Microsoft brings frontier open models powered by Baseten to Microsoft Foundry. Organizations now have more choice across open and closed models, so they can find the right balance of intelligence, performance, and cost for each task. Teams can use these models alongside the models they already use in Foundry, governed by the same enterprise controls and billed under their existing Microsoft agreement.

New model releases keep expanding what’s possible for coding agents, research workflows, and customer support applications. Teams need a practical way to evaluate and adopt open models without sourcing GPU capacity or managing model runtimes. Baseten on Microsoft Foundry handles the serving infrastructure, so teams can focus on finding the right model for each application and putting it into production on the same Foundry platform they use to build and govern AI apps and agents.

"The next wave of AI innovation will come from what developers build when powerful open models meet high-performance infrastructure. Bringing Baseten to Microsoft Foundry gives enterprises more choice and pairs Baseten’s optimized and reliable inference with the governance, controls and observability within the Foundry platform so developers can turn their most ambitious ideas into production applications. Together, we’re helping customers create value faster."

— Steve Sweetman, VP of Product, Microsoft Foundry

Fast, reliable open models with enterprise controls built in

We’re bringing the same performance research and inference infrastructure behind Baseten Model APIs to Microsoft Foundry. Baseten brings research-backed performance and reliability. Microsoft Foundry brings the enterprise security, observability, and billing Azure teams already rely on.

  • Frontier open models: Launch models include DeepSeek V4.1 Flash, GLM-5.3, GLM-5.3-Flash, and Kimi K3, with new models becoming available the day they’re released.

  • Full-stack performance optimization: Every model is tuned at each layer of the stack, from routing and batching to caching and kernels, so applications feel fast to users and stay cost-effective at scale.

  • Production-grade reliability: Infrastructure is designed for 99.9% uptime.

  • Standard and Provisioned deployments: Use Global Standard or Data Zone Standard (US) deployments with pay-per-token billing to evaluate and scale, or Global Provisioned or Data Zone Provisioned (US) deployments with provisioned throughput units (PTUs) for reserved capacity and predictable throughput.

  • Enterprise security and controls: Zero data retention with stateless requests, Microsoft Entra ID and key-based authentication through Foundry, and per-model access controls with Azure Policy.

  • MACC-eligible spend: Usage is billed through Azure as first-party Foundry consumption and counts 100% toward your Microsoft Azure Consumption Commitment (MACC).

Models powered by Baseten are available to all customers in the Foundry model catalog today. An admin opts in to enable access for their organization. Developers then choose a deployment type (Standard or Provisioned), and deployments are ready in seconds. Models are served through the Chat Completions API and the Responses API, so existing OpenAI-compatible code and agent workflows carry over with minimal changes. Requests are stateless, and Baseten stores no prompts or responses.

What's next

Public preview marks the first phase of Baseten on Microsoft Foundry, with general availability expected in early 2027. We’ll keep adding frontier open models as they’re released. Our partnership with Microsoft is another step toward making open models a first-class option wherever enterprises build, pairing Baseten’s fast, reliable inference with the governance and scale of Microsoft Foundry.

To get started, find Baseten in the Microsoft Foundry model catalog and opt in to the preview.