776 of the 388,317 public repositories in the corpus declare a dependency on Pydantic AI. That is 2.7% of the Agent Frameworks category, where it ranks 6 of 27.
Repos are counted by creation quarter, so recent quarters keep growing as the corpus is re-observed. This is adoption among repos we have scanned, not the whole of GitHub.
| LangGraph | 15,902 |
| Claude Agent SDK | 3,043 |
| CrewAI | 2,148 |
| OpenAI Agents | 1,310 |
| Google ADK | 1,283 |
| Pydantic AI | 776 |
| Cloudflare Agents | 616 |
| AutoGen | 607 |
| ComposioHQ/composio | 30,780 | framework |
| pydantic/pydantic-ai | 20,916 | framework |
| Canner/WrenAI | 17,993 | framework |
| Upsonic/Upsonic | 7,956 | framework |
| PrefectHQ/marvin | 6,207 | agent |
| timescale/pgai | 5,806 | agent |
| vitali87/code-graph-rag | 5,290 | framework |
| getzep/zep | 4,966 | agent |
| lemony-ai/cascadeflow | 4,001 | framework |
| hexo-ai/sia | 2,135 | framework |
Detected from declared dependencies in package.json, pyproject.toml and requirements.txt. A project using Pydantic AI without declaring it is not counted, so this is a floor. How this is measured ยท Browse all repos