AutoGen
Microsoft's multi-agent framework (succeeded by Microsoft Agent Framework)
Our Take
AutoGen pioneered the multi-agent conversation pattern. The v0.4 rewrite modernized the architecture significantly (note: AG2 is a separate community fork, not the same as AutoGen v0.4). However, AutoGen is now in maintenance mode — Microsoft is merging it with Semantic Kernel into a unified Microsoft Agent Framework. Still useful for research, but evaluate long-term viability before building production systems on it.
Pros
- + Strong multi-agent conversation patterns
- + Built-in code execution sandbox
- + Backed by Microsoft Research
- + Flexible agent topologies
- + Good for research and prototyping
Cons
- - Now in maintenance mode (merging into Microsoft Agent Framework)
- - API has changed significantly between versions
- - Documentation can lag behind releases
- - No managed cloud offering
Details
- Pricing Model
- open-source
- Starting Price
- $0
- Self-Hosted
- Yes
- Cloud Hosted
- No
- Founded
- 2023
- Repository
- GitHub →
Best For
- • Multi-agent conversations
- • Code generation agents
- • Research and experimentation
- • Enterprise AI workflows
Integrations
OpenAI Azure OpenAI Anthropic .NET Local models
Articles featuring AutoGen
Last updated: