n8n Review 2026: Features, Pricing, Verdict

Updated 2026-09-28 · 1313 words

This page may contain affiliate links. If you buy through them we may earn a commission at no extra cost to you.

Verdict: n8n is the strongest automation platform available if you're willing to either self-host or pay for its cloud plans, and it's the clear pick if AI agents and MCP tooling are part of your roadmap. It asks more of a first-time user than Zapier or Make, and that's the honest trade: more power, more control, and a real learning curve to get there. We ran five workflows on it over two weeks, self-hosted on a small VPS and separately on n8n Cloud, to see where that trade actually pays off.

What n8n actually is

n8n is an open-source, node-based workflow automation platform, self-hostable first and cloud-hosted second. You build a flow by dragging nodes onto a canvas and connecting them with wires that can branch, loop, and handle errors visually, the same general model as Zapier or Make. What sets it apart is that every node can be extended with a full JavaScript or Python code step, and the entire platform runs on your own server for free under n8n's Sustainable Use License if you'd rather not pay for hosting at all. It also ships dedicated AI Agent and LangChain nodes plus native MCP client and trigger support, letting a workflow call external AI tools and expose itself as a callable tool to other agents, all inside the same canvas you use for ordinary automation logic. The January 2026 n8n 2.0 release expanded that library to more than 70 dedicated AI nodes with LangChain treated as a first-class citizen, not a bolted-on feature.

Setting it up

We ran two parallel setups: n8n Cloud for a same-day start, and a self-hosted instance on a $10/month VPS via Docker Compose with Postgres. Cloud was live in minutes, no different from signing up for any SaaS tool. The self-hosted path took under an hour once we'd set the right environment variables for the webhook URL and protocol, the single step most likely to trip up a first-time setup, since a wrong value there means webhook-triggered nodes fail silently against the container's internal address rather than your public domain.

Building real workflows

Our lead-routing flow (webhook in, CRM lookup, conditional Slack alert) took about fifteen minutes, with one small code node added to reshape a Slack payload into the exact format we wanted, a step a more guided builder would have handled with a built-in formatting block instead. A scheduled weekly report reformatting fifty spreadsheet rows ran with no per-row billing concern at all once self-hosted, since n8n's cost model has nothing resembling Make's per-operation billing; it just runs.

The AI-agent workflow is where n8n separated itself clearly from the closed competitors we've tested. We wired an AI Agent node directly to a web-search tool and a Slack output node, and the agent handled its own tool-calling loop natively inside the canvas, no separate "AI agent product" bolted on top of the regular automation logic. We also tested the MCP trigger by exposing one of our own workflows as a callable tool to an external agent, something neither Zapier nor Make currently offers at this level of native, node-based control.

Key features

Beyond the AI-agent tooling, n8n's expression system lets you reference data from any earlier node with {{ $json.field }}-style syntax, which is fast once you're used to it and a genuine hurdle the first week. The node catalog covers 400-plus built-in integrations plus generic HTTP and webhook nodes that can talk to almost any API with a REST interface, so a missing native integration rarely blocks you outright, it just means writing the HTTP call by hand. Error workflows, retry logic, and execution history are all built in and genuinely useful for debugging a flow that fails intermittently in production, something we leaned on constantly once real traffic started hitting our webhooks.

Where it held up

Cost control is the standout for any team willing to self-host: once the VPS is paid for, there's no per-task, per-operation, or per-execution ceiling pushing you toward a plan upgrade the way volume-based competitors do. The AI-agent and MCP tooling is also genuinely ahead of the closed platforms we've compared it against, and it's wired into the same workflow logic you already use for everything else, not a separate product with its own learning curve.

Where it fell short

The learning curve is real and we won't undersell it. A non-technical teammate who's never seen a JSON structure will hit friction faster here than on Make's guided router-and-module model, and the expression syntax in particular takes real practice before it stops feeling like a foreign language. Self-hosting also means you own uptime, backups, and updates; nothing about n8n's own product experience takes that off your plate, and skipping it means paying for cloud plans instead, which start around $24 a month and climb with execution volume.

Pricing

Self-hosted n8n is free indefinitely for internal business use, and that's the real headline: our second workflow costs nothing beyond the fixed price of the box it runs on. If you'd rather not manage a server, n8n Cloud starts around $24 a month for a starter tier with capped active workflows and execution volume, with higher tiers adding more executions and, further up, SSO and other enterprise controls. Either way, check n8n's current plans directly before committing, since execution caps and tier pricing shift as the product evolves. Our full n8n pricing explained breakdown digs deeper into execution-based billing and the hidden costs of each path.

When NOT to use n8n

Skip n8n if nobody on your team wants to own self-hosting and you're not willing to pay for cloud instead; the setup and expression model ask more of a first-time non-technical user than a fully guided builder. It's also not the right choice if your workflows are simple and low-volume enough that Zapier's or Make's gentler on-ramp outweighs any cost savings you'd get from self-hosting, since the time spent climbing n8n's learning curve has a real cost too.

n8n vs the alternatives

Final verdict

n8n earns its reputation as the most capable option in this category if you're building anything beyond simple trigger-action automations, especially AI-agent workflows, and it's the only one of the major platforms that's genuinely free to run at scale once you own the hosting. It's not the right pick for a team that wants zero infrastructure and zero learning curve; Zapier or Make will get a non-technical team to a working automation faster. For the full three-way breakdown including Make, see our Zapier vs Make vs n8n comparison, our head-to-head Make vs n8n digs deeper into that specific trade-off, and our dedicated n8n vs Zapier comparison isolates just that one matchup in more depth. If cost is the main driver behind this search, our cheaper Zapier alternatives roundup covers n8n alongside several others, and for teams ready to commit to self-hosting, our guide to self-hosting n8n walks through the exact production setup we used here. If something about n8n itself, not automation in general, is the friction, our n8n alternatives roundup matches the specific complaint (learning curve, hosting overhead, licensing) to the tool that actually fixes it. New to the platform and want a head start instead of building from a blank canvas, our best n8n workflow templates roundup covers where to find a working starting point; and if you're weighing n8n specifically against OpenAI's newer agent-building tools, our n8n vs OpenAI AgentKit comparison covers that matchup.

Try n8n

Save this to Pinterest