Microsoft expands Foundry model router at Ignite 2026 as costs climb

What happened
Microsoft expanded its Foundry model router at Ignite 2026 with a pool of the top 50 models, as Gartner warns agentic (AI that carries out multi-step tasks rather than answering one question) AI inference costs may rise fivefold by 2028. Microsoft expands Foundry model router at Ignite 2026 as costs climb ⓘ Microsoft.com Microsoft expanded its Foundry model router at Ignite 2026 with a pool of the top 50 models, as Gartner warns agentic AI inference costs may rise fivefold by 2028. Microsoft is a software development and software company based in Redmond, and its products and services include Azure.
At Ignite 2026, which opened in San Francisco on November 17, the company announced an expanded model router for Microsoft Foundry, its AI development platform formerly known as Azure AI Foundry. The firm says routing a task to an agentic reasoning model raises inference costs by at least five times compared with a basic chatbot interaction.
According to the Ignite 2026 Book of News, the router will now draw on an expanded pool featuring the top 50 models in Foundry, work with AI Gateway, and support more Messages and Responses APIs. Microsoft is putting model routing at the center of its pitch to enterprises worried about AI bills.
The timing reflects a wider cost problem. In August, Gartner predicted that AI inference costs per agentic workflow will increase more than fivefold through 2028, even as per-token prices fall.
As one way to keep spending in check, it recommends inference tiering, which sends simpler requests to cheaper models. What changes with the model router The model router is not new.
Microsoft made it generally available at Ignite 2025, and it already works as a single deployment that assesses each prompt and passes it to a suitable underlying model. Microsoft says this lets customers balance quality, cost, and performance without choosing models manually.