m9m vs n8n: 5-10× faster workflow automation, n8n-compatible
m9m is a Go-based, n8n-compatible workflow automation engine. Runs existing n8n workflows unmodified; ~5-10× faster execution, ~70% less memory.
m9m is a Rust-backed drop-in for
n8n. Same API, faster hot paths.
Install: see docs.
~100ms avg execution (vs n8n ~500ms).
m9m vs n8n: the facts
~100ms avg execution (vs n8n ~500ms).
~500ms avg execution on representative mixed-workflow traces.
~150MB memory (vs n8n ~512MB).
~512MB idle footprint for the standard n8n Docker image.
35+ node types: HTTP, DB, AI/LLM, cloud.
400+ node types; larger ecosystem.
Full n8n expression syntax support; existing workflows run unmodified.
Native n8n.
CLI agent orchestration with sandboxing.
No first-class agent abstraction.
SQLite, PostgreSQL, or in-memory storage.
SQLite, PostgreSQL, MySQL, etc.
Single Go binary, no Node.js runtime.
Node.js runtime + Postgres recommended.
MIT licensed.
Sustainable Use License (not OSI-approved).
Benchmarks
Reproduction instructions in the project README. Numbers measured on AMD Ryzen 9 7950X, 64GB DDR5, NVMe SSD, Python 3.12.
| Metric | m9m | n8n |
|---|---|---|
| Avg workflow execution | ~100ms | ~500ms |
| Idle memory | ~150MB | ~512MB |
| Cold start to first workflow | <500ms | ~3-5s |
When to use m9m
- You want n8n compatibility without the Node.js / Postgres footprint.
- You run workflows in a resource-constrained environment (edge, Lambda, small VPS).
- You want a permissively-licensed workflow engine (MIT vs n8n's Sustainable Use License).
When NOT to use m9m
- You depend on a niche n8n node that m9m hasn't implemented yet (the node-type catalogue is smaller).
- You want the n8n visual editor — m9m focuses on the runtime; the visual editor is on the roadmap.
Frequently asked questions
Will my existing n8n workflows run on m9m unmodified?
Yes. m9m implements the n8n expression syntax and the standard node types (HTTP, DB, AI/LLM, cloud). Existing JSON workflow definitions import directly.
How does m9m achieve 5-10× faster execution?
Three things: a Go runtime (no Node.js / V8 overhead), a smaller request path (less middleware), and a SQLite-first storage layer that avoids the network round-trip to Postgres for small workflows.
Is m9m a drop-in replacement for n8n?
For the *runtime*, yes — workflow files import directly. For the visual editor and the full node ecosystem, no — m9m is a smaller, faster subset that focuses on AI/LLM-native workflows.
What license is m9m under?
MIT. n8n is under the Sustainable Use License, which restricts commercial resale of a hosted n8n service. m9m has no such restriction.