Microsoft expands Foundry model router at Ignite 2026 as costs climb
Microsoft unveiled an expanded Foundry model router at Ignite 2026 to help enterprises curb rising AI inference costs.

Key takeaways
- The router now accesses a pool of the top 50 Foundry models.
- Three routing modes let customers balance cost and quality.
- Decision‑1.1 is generally available and runs more than twice as fast.
- Agent 365 Cost Management preview gives admins usage visibility.
In this article
Expanded model router at Ignite 2026

Microsoft announced an expanded Foundry model router at Ignite 2026, opening a pool of the top 50 models.
The new router works with AI Gateway and supports Messages and Responses APIs, letting customers balance quality, cost and performance.
Routing modes and cost controls
Three routing modes are offered: Balanced, Cost and Quality.
Balanced picks the most cost‑effective model in a narrow quality range; Cost widens it to roughly five to six percent.
Decision‑1.1 is generally available and runs more than twice as fast as its predecessor at the same price.
Gartner warns of rising AI inference costs
Gartner predicts AI inference costs per agentic workflow will climb more than fivefold by 2028.
Routing a task to an agentic model can raise inference costs by at least five times versus a simple chat.
Who is affected and next steps
Enterprises and developers using Microsoft Foundry will see new options to manage AI spending.
The router and Decision‑1.1 cut costs keeping model quality, and admins can set policies via Agent 365 Cost Management preview.
Sources
Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.



