n8n vs Pipedream

Quick verdict

n8n is a fair-code, source-available workflow platform: a visual node canvas plus JS/Python code nodes, free Community self-host or managed n8n Cloud, billed per full workflow execution (not per step). Pipedream is a managed serverless integration runtime for developers: pre-built actions plus real Node.js, Python, Go, or Bash with package-manager access, credit-based compute, and a separate Connect product to embed thousands of app auth flows into your own SaaS or agent.

Pick n8n when you want a visual control plane, optional self-host and data residency, AI agent graphs on your infra, and predictable cost at high multi-step volume. Pick Pipedream when you want zero servers, code-native event glue, fast webhook/API automation, or to ship product integrations via Connect without building OAuth for every SaaS.

They solve overlapping “automate SaaS and APIs” jobs but sell different ownership models. n8n sells execution economics and optional ownership of the runtime. Pipedream sells developer velocity on managed workers and embeddable integrations for your end users.

Side-by-side

What n8n is in 2026

n8n is a workflow automation platform aimed at technical teams: drag nodes on a canvas, branch, loop, call APIs, and drop JS or Python where the UI is not enough. The GitHub project is among the most-starred automation repos (n8n pricing/marketing cites on the order of ~197k stars mid-2026). License is fair-code under the Sustainable Use License — source available and free for internal self-host use, with commercial restrictions. It is not classic OSI “pure open source.” White-labeling n8n, hosting it as multi-tenant n8n-as-a-service, or selling a product whose value is substantially the n8n editor itself generally needs a commercial agreement.

Two production paths matter:

• Community Edition (self-hosted): free software on Docker/npm/K8s; you own TLS, backups, upgrades, and Redis/Postgres if you enable queue mode for scale. Collaboration features like SSO/SAML/LDAP, Git environments, and multi-project admin sit on paid Business/Enterprise.
• n8n Cloud: managed hosting. Starter and Pro sell hosted executions with concurrency and project limits. Business is positioned as a self-hosted license with enterprise collaboration features (SSO, Git, environments, scaling options). Enterprise is custom executions, SLA support, advanced security (external secrets, log streaming, long retention).

Product strengths builders cite repeatedly: re-run and debug single steps, pin data, HTTP Request for any API, multi-agent style AI graphs, and “one execution = whole run” economics versus step-tax tools. The integration catalog is smaller than Pipedream’s 3,000+ app marketing number; custom HTTP and community nodes close the gap for engineers who can wire APIs.

Cloud hosting stores data in the EU (Frankfurt). Self-host means you choose region and stack. Business/Enterprise license keys ping n8n’s license server for usage tracking on self-hosted instances — telemetry can be disabled for some data, but production license compliance is still a operational concern for regulated teams.

What Pipedream is in 2026

Pipedream is a serverless integration and compute platform: workflows are a trigger plus ordered steps — either pre-built actions or custom code in Node.js, Python, Golang, or Bash with package-manager access. There is no official self-hosted “run Pipedream forever on your laptop” product. You build on their managed workers (Firecracker-isolated), with SOC 2 Type 2 reports available on request and HIPAA options (BAA) for customers who need to pass PHI.

Two product surfaces:

• Workflows: event-driven automations (HTTP, schedules, app triggers). Credits charge for compute time and memory per workflow segment. Public-registry event sources attached to workflows are documented as free for the source run; development and testing in the builder does not burn credits. Paid plans can run unlimited credits with additional usage charges and optional usage caps.
• Connect: SDKs and APIs so your product can offer end-user OAuth and 3,000+ integrations without building each connector. Connect pricing is documented as two inputs: API usage credits (actions, tool calls, trigger emits to your webhooks, proxy requests) and unique external users who connect accounts. Listing apps/actions and management operations do not consume credits.

GitHub Sync (on higher tiers), dedicated/warm workers against cold starts, Data Stores, and a large public component monorepo on GitHub are the developer-platform story. Pipedream’s components are source-available; the managed runtime is not something you operate like n8n Community.

Pricing and real cost (TCO)

n8n cash cost (official Cloud, annual billing as listed mid-2026 on n8n.io/pricing):

• Community self-host: $0 software; infra often a few dollars to tens per month depending on VPS size and HA — labor is the hidden line item.
• Starter: 20€/mo billed annually — 2,500 workflow executions, unlimited steps per run, 1 shared project, 5 concurrent executions, ~2,300 AI Assistant credits/month, forum support.
• Pro: 50€/mo billed annually — 10,000 executions, 3 shared projects, 20 concurrent executions, admin roles, workflow history, execution search, larger AI credit pool (up to ~13,700/month depending on plan size).
• Business: 667€/mo billed annually — 40,000 executions, self-hosted license features (SSO/SAML/LDAP, Git version control, environments, scaling options). Forum support, not dedicated SLA support.
• Enterprise: custom executions, hosted or self-hosted, 200+ concurrent executions, external secret stores, log streaming, dedicated support with SLA.
• Startup path: under-20-employee companies may qualify for ~50% off Business.

Critical: one execution = one full run, independent of how many nodes fire. Overage framing for Business is documented as extra buckets (e.g. 4,000 EUR for 300,000 executions in n8n’s pricing FAQ — re-check before budgeting). Monthly billing is available but annual is the public headline discount.

Pipedream cash cost (docs + independent 2026 writeups; always re-check pipedream.com/pricing):

• Free: $0 — daily free-credit limit, limited active workflows and connected accounts; unused free credits do not roll over. Independent guides often summarize ~100 credits-scale free capacity and ~3 active workflows.
• Basic: commonly cited around $29/month with a block of included credits (often ~2,000) and higher workflow/account limits.
• Advanced: higher platform fee (independent reports vary, often ~$49–$79 depending on source/era) — GitHub sync, longer history, more capacity features.
• Connect / Business / Enterprise: production Connect, external users, VPC/SLA — custom or higher listed tiers. Community threads cite Connect around ~$99/mo class with included external users and overage per active user (verify current pricing).

Critical credit math from Pipedream docs:

• 1 credit ≈ 30 seconds of compute at default 256MB per workflow segment.
• Double memory to 512MB ≈ double credit burn for the same duration; 1GB ≈ 4×.
• Delays and branches create multiple segments — a short branchy workflow can cost several credits even when wall-clock time is low.
• Most simple linear runs under 30s at 256MB cost 1 credit; steps themselves are not the bill unit (unlike Zapier-style task taxes).
• Paid tiers: unlimited credits with usage charges; you can set usage caps.

TCO people forget:

• n8n self-host: patches, TLS, execution DB growth, queue mode (Redis/Postgres), secret rotation, community node bit-rot, license-server pings for paid self-host features
• n8n Cloud: execution and concurrency ceilings; saved-execution retention caps; AI Assistant credit pools that do not roll over
• Pipedream Free: daily credit walls that stop production workflows mid-day
• Pipedream paid: cold starts unless warm/dedicated workers; credit multipliers for memory and multi-segment flows; Connect external-user counts
• Both: LLM API spend outside the automation bill, retry storms, dual-running during migration, observability tooling

Community sentiment (Reddit, HN, reviews)

n8n praise: Preferred over Make, Pipedream, and Zapier on Hacker News for local unlimited runs, JS/Python code nodes, and strong AI nodes. Reddit technical users call it the right tool when you can self-host and own complexity. Series funding coverage put it firmly in mainstream automation for technical teams. Builders like step-level debug and not paying per node hop.

n8n complaints: Learning curve (JSON, expressions). Self-host is a second job — OOM kills on large payloads (and when the container dies, built-in error workflows may not fire). One bad workflow can lock a single-process instance; queue mode isolates workers but is hard to configure on some PaaS setups (Coolify threads, etc.). Fair-code confuses buyers who wanted AGPL-style freedom. Pure no-code users bounce. License questions about embedding n8n behind a product UI show up regularly on the community forum.

Pipedream praise: Fast for developer event glue, npm in the workflow, managed scale, and Connect for shipping integrations inside a product. Independent reviews frame it as the developer-speed pick versus n8n’s control pick. Auth management for connected accounts is frequently called easier than DIY.

Pipedream complaints: Free-tier credit limits feel tighter than “generous” marketing; community threads call daily credit cliffs a trust hit. Credit math (memory, segments, long jobs) is harder to budget than n8n’s flat execution. “Feels pricey” for agentic/vibe-coding loops that re-run often. Not self-hostable for air-gapped or strict residency shops. Non-coders hit a wall past pre-built actions quickly — Reddit threads comparing n8n vs Pipedream often say “learn n8n if you want a visual system; Pipedream if you already live in code.”

Direct framing: Shopify/marketing threads tend to push Zapier for non-coders, n8n for technical ownership, Pipedream for code-first Shopify/event glue. DevOps automation threads split on whether auth convenience (Pipedream) beats long-term self-host economics (n8n).

When n8n wins

• High multi-step volume where credit/time billing would dominate the monthly bill
• Visual ops + code — ops and engineers share one canvas without every change being a PR to a cloud function
• Self-host / VPC / data residency requirements (you pick region and network path)
• AI agent graphs with tools, memory, and branching as first-class workflow nodes, including local/self-hosted models
• Internal APIs and databases as a long-lived automation backbone
• You already run containers and will own monitoring, backups, and upgrades
• You want execution cost to stay flat when workflows get longer (more nodes), not more expensive per hop

When Pipedream wins

• Serverless default — no Docker, no queue-mode Redis homework
• Code-native steps with full npm/PyPI and real languages beyond a sandboxed snippet
• Product integrations via Connect (end-user OAuth + thousands of apps in your SaaS or agent)
• Burst event glue — webhooks, SaaS triggers, short transforms at low ops cost
• You want GitHub-synced workflows and managed history without standing up n8n Business features yourself
• Team will not operate a self-hosted automation runtime after install day
• You need managed compliance artifacts (SOC 2 report path, HIPAA BAA path) without running your own stack

Risks and failure modes

• n8n “free forever” myth: software free, production not free — disk, backups, exposed instances, unowned upgrades, and 3 a.m. OOM incidents
• n8n license surprises: fair-code blocks multi-tenant resale and n8n-as-a-service without commercial terms
• n8n Community collab gap: multi-user SSO/RBAC and Git environments need paid tiers
• n8n scale path: single-process works until a bad workflow takes the whole instance; queue mode needs Redis/Postgres discipline
• n8n Cloud ceilings: concurrent executions, saved-execution retention, and AI Assistant credit pools can surprise heavy users
• Pipedream free-tier cliff: daily credit exhaustion stops real workloads mid-day
• Pipedream credit surprise: high memory + long jobs + multi-segment control flow multiplies burn
• Pipedream cold starts: latency after idle unless warm/dedicated workers (and their cost)
• Pipedream residency: no self-host product for air-gapped or “data never leaves our VPC” mandates
• Lock-in of process: n8n exports workflow JSON; Pipedream logic lives in their steps/components — both still cost rebuild hours if you leave

Recommendation by profile

Switching notes

From Pipedream to n8n: Re-express steps as nodes; rewrite Node/Python where Pipedream used packages natively. Recreate connected accounts and webhooks. Re-model cost as executions not credits. Plan self-host ops or Cloud plan size from execution logs.

From n8n to Pipedream: Collapse visual graphs into code-first steps; decide which flows become Connect product features vs internal Workflows. Watch memory settings. Expect daily Free limits if you under-provision. Export critical logic and secrets carefully — do not leave live dual-writes without dedupe.

Neither migration is “one-click.” Budget integration re-auth and failure-path retesting, not just happy-path demos.

FAQ

Is n8n free?
Self-hosted Community Edition software is free; you pay for infrastructure and your time. n8n Cloud starts at about 20€/month billed annually for 2,500 executions. Business/Enterprise add collaboration and advanced features.

Is Pipedream free?
There is a permanent Free plan with daily credit limits and caps on active workflows/accounts. Production volume almost always needs Basic or higher. Unused free credits do not roll over.

Is n8n open source?
n8n is fair-code under the Sustainable Use License: source-available and free for internal use, with commercial restrictions. It is not classic OSI open source. Consulting on n8n for clients is generally allowed; reselling n8n-as-a-service is not without a commercial deal.

Can I self-host Pipedream?
Not as a first-class product like n8n Community. Pipedream is managed serverless; components are source-available on GitHub, but the runtime is theirs.

How does pricing really differ?
n8n counts one execution per full workflow run regardless of steps. Pipedream counts credits from compute time and memory per segment (and Connect may also bill external users).

Which is better for AI agents?
n8n if you want visual agent/tool graphs and optional self-hosted models. Pipedream if agents live in your app and need Connect-style access to thousands of user-connected APIs.

Which has more integrations?
Pipedream markets 3,000+ apps and a large component registry. n8n has fewer one-click nodes but reaches anything via HTTP/GraphQL and custom nodes.

Does n8n Cloud keep data in the EU?
n8n documents Cloud data storage in Frankfurt, Germany. Self-host is wherever you deploy.

Does Pipedream offer SOC 2 / HIPAA?
Pipedream documents SOC 2 Type 2 reports on request and HIPAA BAAs for eligible customers. Confirm current scope with their support/security team before committing PHI.

When should I pick Pipedream over n8n?
When you refuse to run servers, think in code/events first, or need Connect to embed integrations for your end users.

Sources

This comparison draws on 162 primary and secondary sources: official product, pricing, and docs for both tools; GitHub repos and issues; n8n and Pipedream community forums; Reddit and Hacker News threads (praise and complaints); independent reviews and pricing analyses; and video walkthroughs. Full annotated list: research_cache/n8n-vs-pipedream_sources.json. Pricing and plan limits change — verify on official pages before purchase.

Bottom line

If your automation layer is a long-lived, multi-step system of record — possibly on your own metal, with agents and humans editing a canvas — choose n8n. If your automation layer is serverless code on events, or you are building a product that must connect each customer’s SaaS accounts, choose Pipedream (Workflows and/or Connect). Many technical teams use both: Pipedream for edge glue and product integrations, n8n for core internal volume and self-hosted AI workflows.

Decide with a 30-day test: (1) pick your three highest-volume flows, (2) estimate n8n executions vs Pipedream credits under real duration/memory, (3) decide whether self-host ops hours are cheaper than managed credits. The wrong pick is usually “we assumed free” — free Community software still costs ops, and free Pipedream still hits daily walls.