Langflow vs n8n (2026)

Components are organized as Core components grouped by purpose and Bundles grouped by service provider, covering language models, embeddings, vector stores, document loaders, data sources and tools.

An API Request component calls arbitrary endpoints, custom Python components extend the catalog arbitrarily, and Tool Mode on any component exposes it to an Agent component as a callable tool. As an MCP client Langflow connects to external MCP servers and uses their tools.

Provider bundles include Apify actors and search providers, and vector database and model provider bundles span the major commercial and local options.

Sourcedocs.langflow.org components overview, bundles, MCP client and API request pagesread 2026-08-31

Several hundred built in app nodes span CRM, marketing, ITSM, HR, databases, warehouses, cloud infrastructure, storage, messaging, security and finance, and major nodes such as Gmail, Google Drive, Google Sheets, Postgres and OpenAI each cover several families of operations. Core nodes add HTTP Request, GraphQL, SSH, Execute Command and Code for anything not in the catalog, community and custom nodes extend the set further, and the MCP Client node connects agents to external MCP servers. Sub-workflows and app nodes can be exposed to agents as tools.

Sourcedocs.n8n.io app nodes, core nodes, MCP Client and how tools work pagesread 2026-08-31

Flows are node graphs of typed components with loop and branching behavior, parsing and type conversion between incompatible ports, and dynamic ports that open as templates reference new variables. Components can be grouped into a single reusable component and saved to the component menu, and freezing a component pins it and all upstream components so their output is reused rather than recomputed. Agents can use components and flows as tools, so one agent can invoke another flow, and individual components can be run in isolation to test dependencies.

Sourcedocs.langflow.org agents, components overview and flows pagesread 2026-08-31

n8n workflows split with If and Switch conditionals, merge data, loop, wait, handle errors gracefully and run workflows with several branches in a defined execution order. Workflows can call other workflows, and converting nodes to a sub-workflow automatically brings along attached AI sub-nodes such as chat models and memory. Agent nodes act as root nodes orchestrating typed sub-node connections for language models, tools, memory and output parsers, and sub-workflows can be attached to an agent as tools. Canvas groups, sticky notes and workflow tags help organize large workflows.

Sourcedocs.n8n.io flow logic tree, break workflows into smaller parts, convert to sub-workflows and how tools work pagesread 2026-08-31

A Webhook component gives each flow its own POST /v1/webhook endpoint, authenticated by API key by default, so an event in an external application starts a run, and flows can also run through the API. Channels include the REST API, an embeddable langflow-chat web component, the Playground and flows published as MCP tools. Langflow has no built-in scheduler or cron trigger.

Sourcedocs.langflow.org llms.txt index and sitemap (webhook and component-webhook pages)read 2026-09-29

Trigger nodes include Schedule Trigger, Webhook, Chat Trigger, Email Trigger over IMAP, RSS Feed Trigger, Local File Trigger, SSE Trigger, n8n Form Trigger, Error Trigger, Activation Trigger, Workflow Trigger, Execute Sub-workflow Trigger, n8n Trigger, Evaluation Trigger, MCP Server Trigger and Microsoft Agent 365 Trigger.

On self hosted instances a durable scheduler runs time based workflows from a queue backed by the database. Agents can be published so people reach them through chat, channels and schedules, and a Respond to Webhook node plus streaming responses complete the request and response surface.

Sourcedocs.n8n.io core node reference, durable scheduler, build and manage agents and stream real-time responses pagesread 2026-08-31

Retrieval is assembled from component classes, document loaders, text splitters such as the Recursive Character Text Splitter with configurable chunk size and overlap, embedding model components, and vector store components supplied as provider bundles, so the customer chooses and owns the store. Retrieval components feed context into prompt templates through typed ports, with a Type Convert component bridging incompatible types. A file management system stores the names, paths, sizes and storage providers of uploaded files, and chat memory can itself be backed by an external vector store.

Sourcedocs.langflow.org components overview, vector store bundles, memory management and file management pagesread 2026-08-31

Data can be stored and searched with vectors through vector databases, embeddings, retrievers and document loaders, and retrieval augmented generation pulls relevant context into a workflow. Vector store providers attach as sub-nodes so the index can be held in a database the customer runs, and a worked example scrapes website content into Pinecone and queries it from a chat workflow. Data tables manage structured tabular data within project boundaries, and Default Data Loader and text splitter components handle ingestion.

Sourcedocs.n8n.io store and search data with vectors, retrieve relevant context, use website content and data tables pagesread 2026-08-31

Memory bases store an agent's long-term chat history in a vector store for semantic retrieval of past conversations across sessions, filling themselves from each attached flow's runs through auto-capture and a configurable ingestion threshold, one memory base per flow and several per flow allowed, on Chroma, Chroma Cloud, OpenSearch or pgvector. Chat conversations are also stored by session ID, with the Agent component's chat memory on by default and a Message History component with store and retrieve modes, and message logs are editable and deletable.

Sourcedocs.langflow.org memory-bases and session-id pagesread 2026-09-29

A memory sub-node on the Tools Agent lets users hold an ongoing conversation across several queries, but that memory does not persist between sessions, and the default Simple Memory node, previously named Window Buffer Memory, is a bounded buffer. Memory attaches to agents as a typed AiMemory sub-node, with session IDs taken from input data or a custom key and optionally scoped per node so separate memory nodes in one workflow do not share history. External chat memory backends can be attached in place of the simple buffer, and Data tables provide structured storage across runs.

Sourcedocs.n8n.io Tools Agent, how memory works, AI Agent common issues and data tables pagesread 2026-08-31

Human-in-the-Loop pauses a flow, creates a checkpoint and waits for a human decision, after which the run resumes from the checkpoint along the selected branch without re-executing completed steps. It works two ways. The Human Input component pauses the flow where it is placed and creates one output branch per configured user action such as Approve or Reject, and per-tool approval on the Agent component, set by enabling Requires approval on an individual tool, pauses the run only when the agent attempts to call that tool. Human-in-the-Loop arrived in the Langflow 1.11 release.

Sourcedocs.langflow.org human-in-the-loop, human-input component and agent tools pagesread 2026-08-31

An AI Agent can be required to get a person's approval before it runs specific tools, and a human fallback brings in a person to answer when the AI cannot help. A dedicated Guardrails node ships as a core node for enforcing policy on AI workflows. Workflow reviews let a builder submit a version for review before publishing, compare changes in a visual diff, discuss them in activity comments, and approve or request changes. Wait and Stop And Error nodes pause and halt runs, and publishing a workflow is a separate step from saving it.

Sourcedocs.n8n.io human-in-the-loop for tools, set a human fallback, Guardrails node, workflow reviews and save and publish pagesread 2026-08-31

Langflow supports API key and JWT authentication, external authentication through an upstream OIDC proxy or corporate SSO gateway with JWKS validation and automatic user provisioning, and a pluggable authorization layer with viewer, developer and admin roles, shares and an authorization audit log. The open-source edition registers a pass-through authorization service and ships no enforcement plugin, so per-resource RBAC is not enforced without a third-party plugin. No security attestation applies to the self-hosted open-source project, and access enforcement is left to the operator.

Sourcedocs.langflow.org authorization and external-authentication pagesread 2026-09-29

An independent auditor evaluates n8n continuously and audits it each year against SOC 2. Its SOC 2 report is available to enterprise customers, a SOC 3 report is published for everyone else, and third party penetration tests of the production environment run at least once a year.

SSO, SAML and LDAP come on the Enterprise plan and advanced RBAC on all paid plans. Security settings add SAML and OIDC single sign on, MFA enforcement, a self service security audit, node blocking, task runner hardening, SSRF protection, encryption key rotation, JWE decryption of OAuth tokens and execution data redaction.

Sourcen8n.io legal security page, docs.n8n.io security configuration treeread 2026-09-29

Each component has an Inspect control showing its output and logs for the most recent run, and a Last Run indicator confirms execution, so a builder can see what every node in the graph produced. Monitor endpoints expose stored messages at GET /v1/monitor/messages, filterable by session ID. The Playground shows message logs per session with full history, and log level and file settings are configurable through environment variables. Langflow does not describe tracing integrations with LangSmith or Langfuse.

Sourcedocs.langflow.org components overview, API reference and playground pagesread 2026-08-31

Executions are first class objects. Teams can view all executions or those of a single workflow, add custom execution data so runs can be filtered by custom fields, and debug by copying earlier execution data into the canvas.

Workflow and node executions can be traced to an OpenTelemetry collector, alongside Prometheus metrics, metrics visualized in Grafana, log configuration and an insights metrics layer.

Workflow change history and version restore are available, execution data can be redacted for compliance, and a self hosted instance can connect to LangSmith for tracing through LangChain in n8n.

Sourcedocs.n8n.io understand executions, trace executions with OpenTelemetry, monitor n8n, view change history and LangChain in n8n pagesread 2026-08-31

Langflow is an MIT-licensed open-source Python application that the customer self-hosts, deployable in containers that serve flows over a network, installable as a Python package, or run as Langflow Desktop, a downloadable application for macOS and Windows that bundles dependencies. LFX builds Langflow as a dependency within another application. Storage is a database the operator controls, LANGFLOW_USE_NOOP_DATABASE disables persistence entirely for testing, and chat memory can be directed to external databases the customer already runs.

Sourcedocs.langflow.org deployment overview, LFX overview, memory management pages and langflow.org desktopread 2026-08-31

n8n installs with a one line setup, Docker, Docker Compose or npm, and deploys to DigitalOcean, Heroku, Hetzner, AWS, Azure, Google Cloud Run, Google Kubernetes and OpenShift Local, alongside managed n8n Cloud. An isolated configuration keeps the instance from connecting with n8n's servers. The customer chooses the database, binary and execution data can go to external storage, queue mode and task runners cover scaling, and an OEM path puts n8n inside another product. Editions run from Community and Registered Community to Business and Enterprise.

Sourcedocs.n8n.io install options, isolate n8n, choose n8n's database, scaling and compare editions pagesread 2026-08-31

Several pre-built templates come ready to use or customize, and many tasks begin by creating a flow from a named template such as Basic Prompting or the Memory Chatbot starter example. New workspaces open with a Starter Project folder by default. Grouped components can be saved into the Core components menu as reusable custom components, and flows are importable and exportable as JSON so complete assets can be shared and adopted.

Sourcedocs.langflow.org get started, memory, web-search and import and export pagesread 2026-08-31

Templates are one of the named ways to build workflows, and self hosted instances can point at a custom workflow templates library. A curated self hosted AI Starter Kit provides a set of AI elements for starting AI workflows quickly. Each node links to browsable template collections for that node, and .n8np packages bundle workflows, folders or whole projects with their dependencies into a portable file for moving between instances. Learning paths offer interactive courses that build real workflows.

Sourcedocs.n8n.io use templates, configure a custom workflow templates library, deploy with the AI starter kit and n8n packages pagesread 2026-08-31

A Language Model component selects provider and model per node, so users add an OpenAI key or pick a different provider and model. Model providers come as Bundles grouped by provider, covering the major commercial options including OpenAI, Anthropic, Google, Cohere and Hugging Face, alongside local models through Ollama. The component can emit either a model response or a Language Model output that powers a downstream component's reasoning, so one flow can drive several models. Custom Python components allow any additional provider to be added.

Sourcedocs.langflow.org components overview, model components and web-search pagesread 2026-08-31

The AI Agent node needs a connected Chat Model sub-node and errors without one, and clicking the Chat Model connector opens a selection of possible chat models. Models attach as sub-nodes, so different agents and chains in one workflow can use different models. Dedicated model provider nodes include OpenAI, Anthropic, Google Gemini, Mistral AI, Perplexity, Qwen Cloud, MiniMax and Moonshot Kimi, and the self hosted AI starter kit runs local models alongside n8n. LangChain is the underlying abstraction, with a LangSmith connection.

Sourcedocs.n8n.io AI Agent node and common issues, what agents do, app node reference and AI starter kitread 2026-08-31

Langflow exposes flows as MCP servers that external MCP clients can use, and it also acts as an MCP client connecting to external MCP servers. A REST API runs flows, lists components at /api/v1/all, triggers runs and returns monitor data, authenticated by API key. Flows import and export as JSON, custom components are written in Python and can be shared, an embeddable langflow-chat web component brings flows into a site with host URL and API key props, and LFX builds Langflow as a dependency.

Sourcedocs.langflow.org MCP, API reference, import and export, custom components and publish pagesread 2026-08-31

The MCP Server Trigger node lets n8n act as a Model Context Protocol server, making n8n tools and workflows available to MCP clients over SSE or streamable HTTP. MCP clients can build and run n8n workflows, and admins oversee connected MCP clients and can revoke their consent. A public REST API is available and can be disabled.

Custom and community nodes extend the catalog, npm modules can be enabled in the Code node, and .n8np packages export workflows, folders or whole projects as portable files with a manifest, size limits and API key scopes. An OEM path embeds n8n in another product using OAuth 2.0 token exchange, iframe SSO and delegated API access.

Sourcedocs.n8n.io MCP Server Trigger, connect to n8n MCP server, n8n packages, security and OEM deployment pagesread 2026-08-31

For testing and debugging, Langflow offers the Playground for interactive flow testing with custom session IDs, running a single component in isolation via build_vertex without executing upstream dependencies, Inspect for per-component output and logs, Freeze to keep a component and its upstream from re-running while preserving last output, message logs that can be edited or deleted to change later chatbot behavior, and automatic backup flows before component updates that may contain breaking changes. Langflow offers no evaluation, scoring, regression or benchmarking capability.

Sourcedocs.langflow.org playground, components overview and memory pagesread 2026-08-31

Evaluation and Evaluation Trigger core nodes ship with the platform. Builders run quick evaluations during development against known test cases and use metric based evaluations to measure, score and improve production workflow performance over time, with guidance on why to test and how to fix common issues. Debugging uses pin and mock data, debug executions that copy earlier execution data into the current workflow, manual and partial runs, and dirty node tracking.

Sourcedocs.n8n.io test and improve AI workflows section, Evaluation and Evaluation Trigger nodes, understand executions pagesread 2026-08-31

No browser, desktop or computer-use component ships with Langflow. The closest capability is the Web Search component, which combines web, news and RSS modes and uses web scraping subject to rate limits, alongside an API Request component for arbitrary HTTP calls. Browser capability reaches Langflow only through third parties, an Apify bundle that calls a hosted RAG Web Browser actor, and community-authored custom components wrapping external browser-automation libraries.

Sourcedocs.langflow.org web-search and components referenceread 2026-08-31

n8n ships an Airtop app node that controls a cloud based web browser for querying, scraping and interacting with web pages, with operations to manage browser sessions and windows and to click an element, fill a form, hover, scroll and type, authenticated with Airtop credentials. Airtop is a third party browser engine wired in through a node, not a browser n8n ships itself. None of n8n's core nodes is a browser, desktop or computer use node, and SSH and Execute Command run commands on remote hosts.

Sourcedocs.n8n.io Airtop node pageread 2026-09-29