Foundry Model Router Expands from Two Regions to 28, Refreshing Its Model Pool
Microsoft's Foundry just went global: model router now spans 28 regions, auto-adds Claude Opus and GPT-5.6 to your pool.
Why it matters
Foundry's regional expansion and automatic model-pool refresh lower friction for enterprises deploying multi-model AI workloads globally, but practitioners need to understand how context-window constraints and deprecated-model handling affect their configs.
The key facts
10 to knowFoundry model router expanded from 2 regions to 28 (global standard) and 21 (data zone)
Claude Opus 4.8 and GPT-5.6 added to model pool
Four deprecated models removed
Default deployments auto-receive pool changes; configured subsets require manual opt-in for new models
Effective context window = smallest model in pool (a constraint practitioners must track)
Published August 2026
New models added: Claude Opus 4.8, GPT-5.6
Default deployments receive pool changes automatically; configured subsets do not until manually added
Effective context window equals the smallest model in the pool
Published August 31, 2026
Go to the source
InfoQ AI/MLinfoq.com
Publisher excerpt: Microsoft expanded Foundry's model router from two regions to 28 for global standard and 21 for data zone deployments, while adding Claude Opus 4.8 and GPT-5.6 and removing four deprecated models. Default deployments receive pool changes automatically; configured subsets exclude new models until…