Microsoft's Copilot harness is now in public preview in Foundry. Runtime is free until Dec 1, then sandbox compute is billed by the vCPU-hour. Nine regions, none in the Middle East, and no bring-your-own-model yet. Full breakdown, pros and cons and what it means for WordPress on heshamsaad.com.
MicrosoftMicrosoft Copilot harness in Foundry - free preview runtimeCloud
Harness runtime free until December 1, 2026 (model inference billed separately)
Microsoft Foundry is Microsoft's managed platform for building and running AI agents. The Copilot harness, the multi-model agent runtime behind GitHub Copilot and VS Code, is now in public preview in Foundry Agent Service, and harness runtime usage is free until December 1, 2026; model inference and connected tools are still billed.
LiveNot verified yet
- Who qualifies
- Any Microsoft Foundry customer with an Azure subscription who creates a prompt agent using the Copilot harness
- How to claim
- See official page
- Expiry
- 2026-12-01
Heads up. The free window covers harness runtime only. Model inference and separately billed tools and connected services are still charged. From December 1, 2026 sandbox compute is billed at $0.0994 per vCPU-hour and memory at $0.0118 per GiB-hour.
Posts 1
Official sources
Added Oct 28, 2026 · Updated Oct 28, 2026, 19:37 UTC
More credits and free plans
View all →
