There are many projects in the AI-agent space, at many layers of the stack. These comparisons are written to help you place Linchpin against each one — what they share, where they differ, when to pick which.
Linchpin is the runtime layer. Three services and Postgres on a single VM. HTTP API for agents, environments, sessions. Per-session Docker sandbox. Append-only event log streamed over SSE. OpenRouter or Ollama for models. Apache-2.0.
If the project you are comparing against is a product (Devin, Manus), Linchpin is the open self-hosted layer you build the equivalent on. If it is a library (LangGraph, CrewAI, AutoGen), Linchpin is a complementary deployment layer — the library can run inside a Linchpin agent, or call Linchpin from your service. If it is another runtime / platform (OpenHands, Dify), the differences are about target user and shape.
See the FAQ for short answers and the glossary for term definitions.