Microsoft Research's open-source multi-agent framework, now archived: Microsoft no longer fixes, secures or supports it and points users to Microsoft Agent Framework.
Agent Development comparison
Pricing, pros, cons, and ideal use cases — side by side.
Microsoft Research's open-source multi-agent framework, now archived: Microsoft no longer fixes, secures or supports it and points users to Microsoft Agent Framework.
Microsoft's open-source SDK for integrating LLMs into C#, Python and Java applications, now succeeded by Microsoft Agent Framework (1.0, April 2026).
| Microsoft AutoGen | Semantic Kernel | |
|---|---|---|
| Pricing | FreeAutoGen's code is MIT-licensed (documentation CC-BY-4.0) and free, but its README declares the project archived and no longer maintained: Microsoft provides no bug fixes, security fixes, new features or support, advises against production use, and directs users to Microsoft Agent Framework. As read on the project's GitHub README on 10 November 2026. | FreeSemantic Kernel is MIT-licensed and free, and Microsoft now describes Microsoft Agent Framework as its "enterprise-ready successor" (version 1.0 released 3 April 2026, also MIT). Its successor, Microsoft Agent Framework, can be hosted on Foundry Agent Service: no added charge for Foundry-native prompt and workflow agents; hosted agents bill $0.0994 per vCPU-hour and $0.0118 per GiB-hour in East US, plus model tokens and tool fees (for example Web Search at $14 per 1,000 transactions). As read on the vendor's pricing page on 10 November 2026. |
| Category | Agent Development | Agent Development |
| Ideal for | Engineering teams prototyping multi-agent systemsResearch and innovation groupsTeams already in the Microsoft ecosystem | .NET and Java enterprise development teamsOrgs embedding AI into existing applicationsMicrosoft-ecosystem shops |
Microsoft AutoGen is built around engineering teams prototyping multi-agent systems; Semantic Kernel leans more toward .net and java enterprise development teams. Shortlist the one whose strengths line up with your biggest constraint.
Get one AI workflow a week showing Agent Development in a real stack — what they cost, and where each one breaks.