n8n

Overview

n8n is a fair-code (source-available) workflow automation platform for technical teams. You build multi-step automations on a visual canvas, drop in JavaScript or Python when you need real code, and connect APIs, databases, SaaS apps, and AI models. n8n GmbH (Berlin) ships both a free self-hosted Community Edition and paid n8n Cloud / licensed Business & Enterprise products. The public product claim centers on ~400+ core integrations, a large community node ecosystem, and ~197k GitHub stars on n8n-io/n8n.

Unlike Zapier or Make, n8n bills paid plans by workflow execution (one full run of a workflow), not per step or per task. Complex multi-node graphs still count as a single execution. That model is the main cost argument vs task-based iPaaS tools when workflows are branchy or data-heavy.

Deployment is the other differentiator: run Community Edition on your own Docker/K8s/VPS with no license fee, buy Cloud Starter/Pro for managed hosting, or license Business (self-host only on the public page) / Enterprise (cloud or self-host) for SSO, Git environments, scaling controls, and support SLAs.

Key features

• Visual workflow editor — Node graph with inspectable input/output at each step; re-run single nodes; pin/mock data for debugging without re-hitting external APIs.
• Code when you need it — Inline JavaScript or Python Code nodes; HTTP Request / GraphQL for anything without a dedicated node; import cURL; custom nodes on self-hosted.
• Integrations — Large first-party node library (CRM, messaging, DBs, cloud, AI). GitHub marketing also cites 1,500+ integrations counting community/community-adjacent packages—verify the nodes you need before migrating production.
• Triggers — Webhooks, schedules/cron, app event triggers, queues; API and CLI control for ops automation.
• AI workflows — Native AI nodes and agent-style patterns; Self-hosted AI Starter Kit (Docker Compose template with local model stack). Cloud plans include AI Assistant credits in preview (monthly allotment; unused credits do not roll over).
• Execution model — Pay (on paid plans) per full workflow run, not per step. Concurrent execution caps differ by plan (e.g. 5 on Starter, 20 on Pro, 200+ on Enterprise per public pricing matrix).
• Collaboration (paid tiers) — Shared projects, admin roles, global variables, workflow history, execution search; RBAC depth increases toward Business/Enterprise.
• Enterprise ops — Git version control, multiple environments, SSO (SAML/LDAP), external secrets, log streaming, queue/worker scaling, insights dashboards with longer retention on higher plans.
• Self-host stack — Docker, npm, and Kubernetes install paths documented; Community Edition is the default without a license key; Business/Enterprise keys unlock gated features and must phone home for license validation on those tiers.

Pricing

Figures below follow the public n8n.io/pricing page (EUR, annual billing called out as “Save 17%” vs monthly). Verified mid-July 2026. Always re-check checkout for your currency, tax, and monthly rates.

Plan	Price (annual billing)	Hosting	Executions / highlights
Community Edition	€0 license (you pay infra)	Self-hosted	Unlimited executions; core product without Business/Enterprise-only features (no SSO, Git environments, shared projects, etc.). Free license key for registered Community unlocks folders, debug-in-editor, custom execution data.
Starter	€20/mo billed annually	Hosted by n8n (Cloud)	2,500 executions/mo; 1 shared project; 5 concurrent executions; unlimited users; ~2,300 AI Assistant credits/mo; forum support
Pro	€50/mo billed annually	Hosted by n8n (Cloud)	10,000 executions/mo; 3 shared projects; 20 concurrent; admin roles; global variables; workflow history; execution search; up to ~13,700 AI credits/mo (plan-size dependent); 7 days of insights
Business	€667/mo billed annually	Self-hosted only (public page)	40,000 executions/mo; 6 shared projects; SSO SAML/LDAP; Git version control; multiple environments; scaling options; 30 days insights; forum support (not dedicated SLA)
Start-up (Business discount)	€333/mo billed annually (50% off Business)	Self-hosted	Same Business feature set/40k executions if eligible: <20 employees, <€5M funding; typically one year (or while still eligible)
Enterprise	Contact sales	Hosted by n8n or self-hosted	Custom execution volume; unlimited shared projects; 200+ concurrent; 365-day insights; external secrets; log streaming; extended retention; dedicated support with SLA; invoice billing

How executions work (official definition)

An execution is one start-to-finish run of a workflow. Step count and data volume do not multiply the bill the way Zapier “tasks” or Make “operations” do. High-frequency schedules still burn quota fast: a workflow every 5 minutes is on the order of ~8,600–8,900 executions/month even if each run is simple.

Business overages (public FAQ)

Workflows keep running past quota; overages may be invoiced. Public Business FAQ states overage buckets of €4,000 for 300,000 extra executions. License keys for Business/Enterprise can apply to unlimited instances, with usage summed across them. Keys ping n8n’s license server daily for paid self-hosted tiers.

Cloud resource notes (trials)

Starter/Pro free trials advertise Pro features with Starter-scale limits (e.g. 5 concurrent, ~320MiB RAM, 2,500 executions, 800 AI credits). No credit card for Starter/Pro trial; Business trial requires a card. Cloud data residency is described as EU (Frankfurt) on the pricing FAQ.

Limits & gotchas

• Cloud execution caps — 2.5k (Starter) / 10k (Pro) / 40k (Business) monthly. Polling or chatty webhooks can exhaust Starter quickly; estimate schedules before committing.
• Concurrency ceilings — 5 / 20 / 200+ concurrent executions by plan. Bursts queue or wait when you exceed concurrent slots.
• Execution retention — Saved execution count, storage (Cloud), and day windows differ by plan (e.g. short retention on lower Cloud tiers). Hitting history limits does not stop runs; it only trims debug history.
• AI Assistant credits — Monthly Cloud allotments; no roll-over; cannot buy extra packs on the public FAQ (upgrade plan instead). Self-hosted AI Assistant is “coming soon” with BYO API keys (no credit meter once available).
• Community Edition gaps — No custom variables, environments, external secrets, log streaming, multi-main, projects, SSO, workflow/credential sharing, or Git version control (per docs community feature list). Queue mode itself is included; multi-main is not.
• License / fair-code — n8n is source-available under a Sustainable Use / fair-code style license, not OSI “open source” in the pure sense. Internal self-host use is the common free path; offering n8n as a commercial service or certain redistribution cases can require a commercial deal—read the license and ask legal if you resell automation.
• Business is self-host-only on the marketing page — Do not assume Cloud Business. Startup 50% is Business-class self-host, not a Cloud discount.
• Support ladder — Forum for most plans; dedicated support with SLA is Enterprise. Business is self-serve with forum only per FAQ.
• Ops tax on self-host — You own upgrades, backups, TLS, DB (Postgres recommended for production), monitoring, and HA. Cheap VPS (~€5–20/mo) is common for hobby/prod-lite; enterprise HA is real DevOps work.
• Enterprise quotes vary — Community posts cite five-figure EUR/year deals; treat third-party numbers as anecdotes only.

Community sentiment

Sentiment splits cleanly along deployment and pricing lines.

Praise: Developers and ops teams like execution-based economics vs Zapier/Make for complex graphs, full data control when self-hosting, and the ability to mix visual nodes with real code. GitHub popularity and G2 (~4.7–4.8) show a strong technical following. Self-host on a small VPS or home lab for personal/internal automation is widely recommended as “almost free.”

Criticism: The jump from free Community to paid Business (~€667/mo annually for 40k executions) drew sharp Reddit/HN backlash when self-hosted Business billed on executions—users argue they already pay for servers. HN threads on fundraising and licensing call out that self-host paid tiers can still throttle “imaginary” execution quotas and that the license is not classic open source. Cloud Starter’s 2.5k executions is frequently called tight for anything beyond light automation. Enterprise price quotes (anecdotal five figures) frustrate teams who assumed free self-host covered production multi-user governance.

Who should use it

• Technical builders who are fine with JSON, APIs, and occasional JS/Python and want more power than pure no-code Zaps.
• Teams leaving Zapier/Make because multi-step task/operation billing got expensive relative to workflow complexity.
• Security-conscious orgs that must keep credentials and payload data on their own infrastructure (Community or licensed self-host).
• AI automation experiments that need agents + tools + human-in-the-loop patterns without a full custom backend.
• Not ideal for non-technical solo marketers who want pure click-and-go with vendor hand-holding, or orgs that need SSO/audit/Git but refuse Business/Enterprise pricing—those features are gated.

Alternatives

• Zapier — Fastest no-code SaaS integrations; task-based pricing; weak self-host story.
• Make — Visual scenarios with strong UX; operations-based pricing; cloud-first.
• Pipedream — Code-first serverless workflows; generous free tier; less traditional self-host.
• Tray.io — Enterprise iPaaS with sales-led pricing for larger orgs.
• ActiveCampaign / CRM suites — Use when automation is mostly marketing journeys, not general iPaaS.
• Retool Workflows — Better when the center of gravity is internal apps + light automation, not pure integration volume.
• Node-RED / custom workers — Fully free OSS or code when you outgrow visual tools or license constraints.

Verdict

n8n remains one of the strongest choices in 2026 for technical workflow automation: real code nodes, execution-not-step economics, and a genuine free self-host path. Official list pricing is clear—Cloud from €20/mo (Starter, annual) and €50/mo (Pro); self-host Business at €667/mo annual for 40k executions; free Community for unlimited runs without the enterprise feature pack; Enterprise and Startup (50% Business) for special cases.

Buy Cloud when managed reliability beats VPS chores and your monthly runs fit the tier. Self-host Community when cost and data control win and you can live without SSO/Git/projects. Only pay Business/Enterprise when compliance, multi-user governance, or SLA support are real requirements—not because you assumed “self-host should include everything free.” Re-read the fair-code license before embedding n8n in a commercial product you resell.