n8n Review (2026)
Visual + code automation for technical teams: JS/Python inline, Docker/K8s self-hosting, 500+ integrations, production AI agents with HITL guardrails. 4.9/5 on G2. Best fit: engineering and ops teams needing code-level control. Watch-out: pricing is execution-based and not publicly listed.
n8n is the top pick for technical teams who need visual workflow orchestration with inline JS/Python code steps and full self-hosting on Docker or Kubernetes. Pricing is execution-based and not publicly listed: verify at n8n.io/pricing before budgeting, and note that no-code teams will hit more friction here than with Make or Zapier.
Audience Fit
n8n: built for technical teams, not no-code operators
Best for
Dev/ops teams: code + visual in one workflow
Self-hosting: Docker, K8s, air-gapped, free
AI agents: HITL, RAG, multi-step, production
Complex workflows: per-execution beats per-step pricing
Skip if
No-code teams: high technical skill floor
App coverage: 500 nodes vs Zapier's 8,000+
Pricing approval: not listed, requires direct verification
Pricing
Execution-based tiers: unlimited users and workflows
All plans include unlimited users and workflows. The variable driving plan selection is monthly executions, not seats. Exact prices are not published: verify at n8n.io/pricing before committing.
| Plan | Workflow history | AI credits/mo | Saved executions | Log retention | Concurrent runs | Shared projects |
|---|---|---|---|---|---|---|
| Starter | 1 day | 50 | 2,500 | 7 days | 5 | 1 |
| Pro | 5 days | 150 | 25,000 | 30 days | 20 | 3 |
| Enterprise | 365+ days | 1,000 | 50,000 | Unlimited | 200+ | Unlimited |
Open-source. Docker or Kubernetes. Unlimited workflows and executions.
- Unlimited workflows, users, executions
- 500+ integrations
- Code steps (JS/Python + npm/pip)
- AI agent building
- Queue mode for high-volume loads
- Full data residency control
- Cloud hosting
- SSO/LDAP (Enterprise cloud)
1-day history · 50 AI credits/mo · 2,500 saved executions · 5 concurrent runs
- Unlimited users and workflows
- 50 AI Workflow Builder credits/mo
- 2,500 saved executions
- 7-day log retention
- 5 concurrent executions
- 1 shared project
- 500+ integrations
5-day history · 150 AI credits/mo · 25,000 saved executions · 20 concurrent runs
- Unlimited users and workflows
- 150 AI Workflow Builder credits/mo
- 25,000 saved executions
- 30-day log retention
- 20 concurrent executions
- 3 shared projects
- SSO/LDAP (Enterprise only)
Tier limits above are confirmed from official documentation. Specific currency amounts were not captured at time of research. All plans include unlimited users and workflows: seat-based pricing is not a factor.
Feature Overview
What n8n has that Make and Zapier don't
The core differentiator is the hybrid model: every workflow combines visual nodes and code steps in the same execution graph, without a parallel codebase or a separate service.
| Capability | What it covers | Availability |
|---|---|---|
| Visual Workflow Editor | Drag-and-drop canvas with 500+ nodes. Triggers, filters, loops, conditionals, scheduling. Inline logs and saved executions for in-editor debugging. | All plans |
| Code Steps (JS/Python) | Drop a Code node anywhere: access upstream data, run JS or Python, pass results downstream. Install npm packages or Python libraries on self-hosted instances. | All plans (npm/pip: self-hosted) |
| Self-Hosting | Docker, Kubernetes, or any container environment including air-gapped networks. Full data residency: workflow data, credentials, and execution history stay in your own infrastructure. | Self-hosted (open source) |
| AI Agent Building | Multi-step agents, multi-agent orchestration, and RAG agents with memory and tool use. Human-in-the-loop guardrails pause execution for human approval before irreversible actions. | All plans |
| Developer Tooling | Import cURL to generate HTTP nodes. GraphQL support. Webhook nodes. Git integration and workflow diff for version control across deployments. | All plans (queue mode: self-hosted) |
| Enterprise Security | SSO (SAML), LDAP, 2FA, encrypted secrets, external secret stores (AWS, Azure, GCP, HashiCorp, Infisical), audit logging with log streaming. | Enterprise |
| n8n Embed | White-label n8n and embed automation into your own product for customer-facing workflow features without building a custom engine. | Enterprise |
Make charges one credit per module per run. A 15-step scenario at 1,000 runs/month costs 15,000 credits. n8n charges per workflow execution: same scenario costs 1,000 executions. For multi-step workflows at high frequency, n8n's model is significantly cheaper.
Strengths vs Tradeoffs
4 real strengths, 4 real tradeoffs
Strengths
G2 rating: 4.9/5, highest in category
Self-host: unlimited free on Docker/K8s
Execution billing: cheaper than per-step at scale
AI agents: HITL, RAG, production-grade
Tradeoffs
Pricing: not listed publicly, verify required
Skill floor: higher than Make or Zapier
App library: 500 vs Zapier's 8,000+
Enterprise SSO: locked to Enterprise tier
Stack Fit
n8n vs Make vs Zapier: which fits your team
Three platforms targeting different buyer profiles. The wrong choice creates a re-platform project six months later.
| Dimension | n8n | Make | Zapier |
|---|---|---|---|
| Pricing model | Per workflow execution | Per module action (per step) | Per task (per Zap run) |
| App library | 500+ native nodes | 3,000+ | 8,000+ |
| Code steps | JS + Python inline | No native code steps | No native code steps |
| Self-hosting | Full: Docker, K8s, air-gapped | Enterprise on-prem only | No |
| AI agents | Multi-step, HITL, RAG | AI Agents (beta) + MCP | Copilot + Zapier MCP |
| Skill floor | Advanced (technical teams) | Intermediate | Beginner |
| Pricing transparency | Not published | Published from $9/mo | Partially published |
| Best for | Engineers: code + self-hosting | Ops: complex branching logic | Teams: broadest app coverage |
Alternatives
Not the right fit? Closest options
FAQs
4 common questions about n8n
Yes if you self-host: Docker or Kubernetes deployment is free with unlimited workflows and executions. Cloud plans are paid with a 14-day free trial; verify current pricing at n8n.io/pricing.
A Code node placed anywhere accepts upstream data, runs JS or Python, and passes results to the next node. Self-hosted instances also support npm packages and Python libraries.
The visual editor covers many use cases without code. Advanced setup, custom HTTP nodes, and AI agents require developer familiarity. Zapier or Make is better for non-technical teams.
HITL guardrails pause agent execution at defined points, requiring human approval before the next action runs. This prevents irreversible errors in production workflows like sending emails or updating CRM records.
Bottom Line
Should you use n8n?
Use n8n if your team writes code, needs Docker/Kubernetes self-hosting for data control, or builds production AI agents where HITL and multi-step reliability matter. Pass if you need a no-code experience or the broadest possible app library: Zapier and Make cover both faster.
Ready to try n8n?
14-day free cloud trial, no credit card. Self-host free on Docker or Kubernetes: unlimited workflows and executions.