n8n Pricing: Self-Hosted vs Cloud

n8n Pricing Model

n8n uses an execution-based pricing model for its cloud offering, where one execution equals one complete workflow run regardless of the number of nodes in the workflow. This is fundamentally different from Zapier's per-task billing, where each step in a workflow counts as a separate task. The distinction matters significantly for AI workflows, which typically involve many nodes (trigger, data processing, LLM call, memory, output formatting, action) in a single execution.

Self-Hosted Community Edition

The Community Edition is completely free with no license fees, no execution limits, no workflow limits, and no user limits. Every feature is included, including all AI nodes, LangChain integration, and the full integration catalog of 500+ connectors. The only cost is your infrastructure.

Infrastructure costs for self-hosting are minimal. A basic VPS from Hetzner (CX22, 2 vCPU, 4GB RAM) costs around 4 euros per month and handles moderate workloads comfortably. For AI workloads with local model inference via Ollama, you need more resources, typically a server with at least 16GB RAM and a GPU if you want reasonable inference speed. A dedicated GPU server from providers like Hetzner or OVH runs 30 to 80 euros per month depending on the GPU tier.

The total cost of ownership for self-hosting includes the server cost plus your time for maintenance. If you are comfortable with Docker and basic Linux administration, expect one to two hours per month for updates, monitoring, and occasional troubleshooting. For teams without DevOps expertise, the maintenance overhead can be significantly higher.

Cloud Plans

n8n Cloud runs on n8n's managed infrastructure with automatic updates, backups, and scaling. The Starter plan costs 24 euros per month (billed annually) and includes 2,500 executions with unlimited workflows and users. The Pro plan costs 60 euros per month for 10,000 executions with additional features like version history. The Enterprise plan starts at 800 euros per month with 40,000 executions, SSO, dedicated infrastructure, and priority support.

All cloud plans count only successful executions. Failed executions and test runs during development do not count toward your monthly limit. This is a genuinely helpful policy during the development phase when you are iterating on workflows and running many test executions.

Overage pricing applies when you exceed your plan's execution limit. Additional executions are billed at a per-execution rate that varies by plan tier. For predictable workloads, you can estimate costs accurately. For variable workloads, monitor usage through the n8n dashboard and set up alerts before hitting your limit.

Enterprise Pricing

Enterprise features are available for both self-hosted and cloud deployments. Self-hosted Enterprise adds SSO via SAML and LDAP, advanced role-based access control with granular permissions, audit logging for compliance, log streaming to external systems, and priority support with SLA guarantees. Cloud Enterprise adds all of the above plus dedicated infrastructure and custom execution limits.

Enterprise pricing is not publicly listed and requires contacting n8n sales. Based on community reports, self-hosted Enterprise starts around 300 euros per month for small teams, with pricing scaling based on the number of users and support level. The pricing restructure in late 2025 moved to tiered plans, which generated mixed reactions in the community.

Cost Comparison with Competitors

The most revealing comparison is at the workflow level. Consider a 10-node AI workflow that runs 5,000 times per month. On n8n Cloud Pro (60 euros/month), that uses 5,000 of your 10,000 executions, leaving room for other workflows. On Zapier, the same workflow would consume 50,000 tasks (10 steps x 5,000 runs), requiring the Professional plan at $99/month or higher.

Make (formerly Integromat) counts operations similarly to Zapier but at lower prices. The Pro plan at $16/month includes 10,000 operations, which would cover 1,000 runs of a 10-step workflow. For 5,000 runs, you would need 50,000 operations, requiring a higher tier at around $50/month.

Self-hosted n8n eliminates execution costs entirely. The same 5,000 monthly runs of a 10-step workflow costs only your server infrastructure, typically $5 to $7 for a basic VPS. For high-volume workloads, the savings compound rapidly. A team running 50,000 executions per month would pay 60 euros on n8n Cloud Pro but potentially hundreds on competing platforms.

AI-Specific Cost Considerations

n8n platform costs are only part of the equation for AI workflows. LLM API costs (OpenAI, Anthropic, etc.) add a per-request expense that n8n does not control. A single GPT-4o call costs roughly $0.005 to $0.02 depending on input/output length. At 5,000 executions per month with one LLM call each, API costs add $25 to $100 monthly.

Self-hosted setups with Ollama eliminate LLM API costs entirely, replacing them with hardware costs. If you already have suitable hardware, the marginal cost of running local inference is essentially zero beyond electricity. This makes self-hosted n8n with Ollama the lowest-cost option for high-volume AI workloads by a significant margin.