Microsoft AFMR (web)

Citation

Microsoft Research. “Accelerating Foundation Models Research.” Program page, archived, date not stated. https://www.microsoft.com/en-us/research/collaboration/accelerating-foundation-models-research/. Read 2026-10-15.

Bottom line

• Microsoft’s Accelerating Foundation Models Research (AFMR) program has ended. The page says the program “has reached its conclusion” and is kept “for reference and archival purposes”.
• AFMR gave access to “state-of-the-art foundation models hosted on Microsoft Azure”. It used “Calls for Proposals”.
• The successor is the Agentic AI Research and Innovation (AARI) initiative. The AARI page lists funded projects but no open call, eligibility or application steps.

Background

AFMR had three goals: align AI with human goals, improve human interactions and accelerate scientific discovery in the natural sciences. The archived page links to collaborator lists and news. It does not keep the old calls’ terms.

Key claims

• Status: “The Accelerating Foundation Models Research program has reached its conclusion. These pages are maintained for reference and archival purposes.” (Note under Highlights)
• Offer: “AFMR includes a grant program that provides access to state-of-the-art foundation models hosted on Microsoft Azure through Microsoft Azure AI services.” (Program description)
• Process: “we engage with the research community through Calls for Proposals, mind-swaps, workshops, conferences.” (Program description)
• Successor: “Microsoft Research is now building on the momentum of this research via The Agentic AI Research and Innovation (AARI) initiative. AARI is not a new beginning, but a natural evolution of the foundation AFMR laid.” (Program description)
• AARI works “through Grand Challenges with the academic community and Microsoft Research.” (Program description)

Evidence and method

The program’s own archived page. That page lists 12 funded projects, each with an academic lead and a Microsoft Research lead. It states no application process.

Limits

• The archived page does not state eligibility, application steps, review, award size, duration, reporting or usage rules.
• Older AFMR calls had these terms on separate call pages. University pages and slides that repeat them are not the operator’s own record.