How much does Latenode cost?

Free forever with 10,000 CPU seconds a month and 5 active workflows; beyond that, pay as you go for runtime from $0.00012 per CPU second down to $0.00005 in volume brackets, with no base fee. The earlier credit plans from $5 a month are no longer on the pricing page.

What are the best alternatives to Latenode?

Based on documented capability coverage across the Agentic Index 14 point taxonomy, the closest alternatives to Latenode among agent builders are Sema4.ai, StackAI, Airtable, Base44. Each is graded on the same public evidence, and no vendor pays for placement.

Does Latenode support MCP and tool calling?

Full under the Agentic Index verification standard. Latenode publishes public evidence of Model Context Protocol (MCP) support, tool calling, and API or SDK extensibility for connecting agents to external tools. Evidence basis: Cursor, Claude or another MCP client can build and run scenarios programmatically through the Latenode MCP Server, which comes with a full tools reference and a tool that runs any single catalog action from an agent without creating a scena. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does Latenode have SOC 2, SSO and role based access control?

Full under the Agentic Index verification standard. Latenode publishes public evidence of enterprise security and identity controls, covering a compliance posture such as SOC 2 or ISO 27001, single sign on (SSO or SAML), and role based access control (RBAC). Evidence basis: SOC 2 Type II certification with single sign on, role based access control and GDPR compliance, published as enterprise grade security and repeated on Latenode's own solution pages. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Can Latenode be self hosted or run in a private VPC?

Full under the Agentic Index verification standard. Latenode publishes public evidence of self hosted or private VPC deployment, along with data residency options for controlling where agent data is processed and stored. Evidence basis: Self hosting is offered for teams prioritizing data security, giving full control over infrastructure and data, and the platform ships across on premise and cloud hosted environments. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Back to vendors

Latenode

Q: Who is Latenode for?

technical operations teams, Zapier and Make switchers at volume, companies embedding white label automation

Visit site

Agent builderindependentVerified 2026-09-29

Entry priceFree with 10,000 CPU seconds a month; then pay as you go from $0.00012 per CPU second, no base feeFull pricing detail

Low code agent and automation platform billing by execution seconds rather than per task, with a built in headless browser and database, bundled access to GPT, Claude and Gemini, and self hosted or on premise deployment.

Latenode is a low code automation and integration platform that has repositioned around building autonomous agents rather than moving data between apps. Its distinguishing commercial choice is the billing unit: it charges for execution time in processing seconds rather than per operation or per task, which is why comparisons put roughly 100,000 operations at about $19 against several hundred dollars on per task platforms.

The technical surface is broad. A visual builder sits over full JavaScript with access to NPM packages, with an AI code copilot generating and optimizing the code. A headless browser is built in, so agents scrape, fill forms and interact with sites that have no API without bolting on a separate service. A built in database stores and queries structured data inside workflows. Model access is bundled, covering GPT, Claude, Gemini and a long tail of others under one subscription rather than requiring separate provider keys, and Latenode's own pages state 5,500+ integrations across 800+ apps. It also documents three distinct agent architectures, tool centric, sub agent and action selection.

Operationally it offers real time monitoring with trace inspection and step by step debugging, webhook driven real time triggers, conditional branching, and a white label embedded iPaaS for companies that want to resell automation inside their own product. Security covers SOC 2 Type II, single sign on, role based access control and GDPR, and it can be self hosted or run on premise.

One caution belongs with all of that: the execution time billing that makes it cheap at volume is hard to forecast, and independent analysis warns that long running steps, meaning AI calls, headless browser sessions and heavy transforms, consume disproportionately more than a run count suggests.

Vendor details

Canonical URL

https://latenode.com

Company status

independent

Use cases & customers

Target customers

technical operations teamsZapier and Make switchers at volumecompanies embedding white label automation

Deployment options

cloudself-hostedon-premise

Integrations

5,500+ integrations across 800+ apps per the vendor's own homepage and documentation, with named connectors spanning CRM (HubSpot, Salesforce), support and ITSM (Zendesk, ServiceNow), documents (Google Drive, Notion), data (PostgreSQL, MongoDB, Snowflake, BigQuery, Pinecone), messaging (Slack, Gmail), development (GitHub, Jira, Linear), payments (Stripe) and commerce (Shopify). MCP-exposed services are reachable through the integrations surface. Beyond the catalog, full JavaScript in any node with over a million NPM packages plus HTTP request support means any library or API is callable from a workflow. The vendor's AI model count is stated inconsistently across its own surfaces, at 1,200+ in the documentation, 335+ in a homepage title and over 200 in a blog post.

In practice

Your per task automation bill scales faster than your usage. Latenode charges for execution seconds instead, which is why comparisons put 100,000 operations near $19 rather than several hundred dollars.

Your agent needs to work against a system with no API. A headless browser is built into the platform, so scraping, form filling and site interaction happen natively rather than through a bolted on service.

You want one subscription instead of separate OpenAI, Anthropic and Google accounts. Model access is bundled, covering GPT, Claude, Gemini and a long tail of others.

Sources & related URLs

Related / legacy domains

https://latenode.com/products/rag https://latenode.com/integrations https://documentation.latenode.com/get-started/introduction https://help.latenode.com/

Research sources

https://latenode.com/ https://latenode.com/pricing-plans https://latenode.com/products/rag https://latenode.com/integrations https://documentation.latenode.com/get-started/introduction https://help.latenode.com/advanced-features/46DTZD5aggy7NR22L329Ge/using-ai-agent-with-rag/5uF81TjKG8k6Yrmm3CvyZD

Agentic Index coverage score

12.5 / 14 capabilities · 89%

Integrations & Tool Calling	Full
The integration catalog runs to 5,500+ integrations across 800+ apps per the homepage, and the documentation introduction states 1,200+ AI models usable without managing API keys alongside 5,500+ integrations. Named connectors span CRM with HubSpot and Salesforce, support and ITSM with Zendesk and ServiceNow, documents with Google Drive and Notion, data with PostgreSQL, MongoDB, Snowflake and BigQuery, messaging with Slack and Gmail, development with GitHub, Jira and Linear, payments with Stripe, commerce with Shopify, plus Airtable and Pinecone. The integrations page covers MCP servers, so MCP-exposed services are reachable as integrations. Full JavaScript in any node with over a million NPM packages and HTTP request support extend reach beyond the catalog. The vendor's own model count varies across surfaces between 1,200-plus, 335-plus and over 200. Sourcelatenode.com home and integrations pages, documentation.latenode.com introductionread 2026-08-31
Workflow Orchestration	Full
Multi agent orchestration with three documented agent architectures, tool centric, sub agent and action selection, alongside conditional logic, branching workflows and iterators for large data sets. Sourcelatenode.com/pricing-plansread 2026-09-29
Knowledge Grounding & RAG	Full
The vendor publishes a dedicated RAG product page stating the engine uses semantic search, vector search, BM25, filters and ranking rules to find relevant information for a query, with ingestion from PDFs, documents, databases, APIs and websites and named connectors including Google Drive, Notion, PostgreSQL, MongoDB, Snowflake, BigQuery, Zendesk and Pinecone. Knowledge is stored on the platform with the vendor stating no external Postgres, vector database or extra connections are required. The help center documents connecting a RAG Search node to an AI Agent node using fromAIAgent() so retrieval is invoked as an agent tool, and the pricing page lists grounding agents in searchable knowledge bases as available from the free tier. No citation or source-attribution behavior in agent answers is documented. Sourcelatenode.com products/rag and rag pages, help.latenode.com AI Agent with RAG documentation, latenode.com home and pricing pagesread 2026-08-31
Human Oversight & Guardrails	Partial
Latenode's oversight is guidance and limits rather than approval. Its guardrails guide recommends embedding rules in the agent's system message, such as requesting confirmation before sensitive actions, limiting agents to focused use cases, connecting only necessary tools and validating inputs and outputs, and the AI Agent node caps tool invocations per run with a max-iterations setting. No approval node or pause for a human decision is documented. The homepage's human approval gates describe what Latenode's done-for-you team builds into delivered automations, not a platform feature. Sourcedocumentation.latenode.com guardrails and ai-agent-node pagesread 2026-09-29
Security, Identity & Governance	Full
SOC 2 Type II certification with single sign on, role based access control and GDPR compliance, published as enterprise grade security and repeated on Latenode's own solution pages. Sourcelatenode.com/pricing-plansread 2026-09-29
Observability & Auditability	Full
Latenode documents an Execution History for viewing, analyzing and debugging past scenario runs, and version history with development and production branches. Think and Plan tools are listed as reasoning tools that improve agent transparency, run history is kept 3 days on Free and 30 days on pay as you go, and the pricing page has described execution tracking that reviews every step an agent took. Sourcedocumentation.latenode.com llms.txt index, latenode.com/pricing-plansread 2026-09-29
Memory & State Persistence	Partial
The AI Agent node keeps short-term memory within a session: a Session ID such as a chat or user ID loads that thread's conversation history into context, with a configurable number of message pairs, and an empty ID starts a new session. A built-in database lets scenarios store and query structured data across executions. No agent memory with a stated lifetime, purge path or review and delete surface is documented. Sourcedocumentation.latenode.com ai-agent-node pageread 2026-09-29
Deployment & Data Residency	Full
Self hosting is offered for teams prioritizing data security, giving full control over infrastructure and data, and the platform ships across on premise and cloud hosted environments. Sourcelatenode.com/pricing-plansread 2026-09-29
Prebuilt Agents, Templates & Packs	Full
A Templates catalog offers ready to use workflows copied into a workspace from a gallery, filterable by topics including AI, agents and chatbots, headless browser, messaging and bots, CRM and sales and documents and invoice processing, each named and described with an estimated run cost, for example an RSS to Discord poster with AI generated posts and images and a travel booking email written by ChatGPT. The docs document copying a scenario from the template gallery and adjusting authorizations. Sourcelatenode.com/templates and documentation.latenode.com template-galleryread 2026-09-29
Triggers & Channel Coverage	Full
A webhook system provides real time triggers so agents respond instantly to external events or user actions, combined with conditional logic and branching so workflows adjust to changing conditions without manual input. Sourcelatenode.com/pricing-plansread 2026-09-29
Model Flexibility & Routing	Full
Native access to top tier language models including OpenAI, Claude, Deepseek and LLaMA comes under "one subscription for multiple LLMs", with GPT, Claude and Gemini named and a long tail of additional models available, removing the need for separate provider accounts. Sourcelatenode.com/pricing-plansread 2026-09-29
APIs, SDKs & MCP Extensibility	Full
Cursor, Claude or another MCP client can build and run scenarios programmatically through the Latenode MCP Server, which comes with a full tools reference and a tool that runs any single catalog action from an agent without creating a scenario. Full JavaScript in any node with NPM packages, HTTP requests, webhooks and a white-label embedded iPaaS extend the platform further. Sourcedocumentation.latenode.com MCP server section and llms.txt indexread 2026-09-29
Testing, Debugging & Optimization	Partial
Detailed execution tracking and debugging tools allow reviewing every step, testing different parameters and tuning performance, and an AI code copilot optimizes generated JavaScript, but no evaluation harness, benchmark suite or regression testing for agent behavior is documented. Sourcelatenode.com/pricing-plansread 2026-09-29
Browser & Computer Use	Full
A Puppeteer powered headless browser is built into the platform, letting agents scrape data, fill out forms and interact with websites natively, explicitly so that systems without APIs can be automated without an external browser service. Sourcelatenode.com/pricing-plansread 2026-09-29

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Pricing

Free with 10,000 CPU seconds a month; then pay as you go from $0.00012 per CPU second, no base fee

CPU seconds of workflow runtime

Free tierTrial available

What is public

Tier structure, the free allowance and the execution time model are published on Latenode's own pricing page and independently verified by a third party cost tracker.

Billing mechanics

Latenode charges for compute time in CPU seconds rather than per operation, node or API call. The free tier includes 10,000 CPU seconds a month and 5 active workflows; beyond that, pay as you go runs from $0.00012 per CPU second down to $0.00005 in volume brackets, with no base fee. Every plan includes AI access and the full integration library rather than gating either behind a tier, and enterprise terms are custom.

Cost watchouts

CPU second metering means long running steps cost disproportionately more than a run count suggests, and AI calls, headless browser sessions and heavy transforms are exactly the steps an agent workload is made of

Variable cost rationale

Time based metering is cheaper than per task billing at volume but harder to predict, and the exposure concentrates in exactly the agentic steps, model calls and browser sessions, that run longest.

Additional watchouts

The cheapness is real but the predictability is not. Independent analysis is explicit that CPU second metering is hard to forecast from run counts alone, and that long running steps cost disproportionately more, naming AI calls, headless browser sessions and heavy transforms. Those are precisely the steps that make a workflow agentic, so the cost advantage narrows as the work becomes more AI heavy. Model a representative workflow rather than extrapolating from a run count.

Overage / add-ons

Runtime beyond the free 10,000 CPU seconds bills per CPU second in brackets from $0.00012 down to $0.00005; the number of steps does not multiply the bill.

Sales call required

No, self serve available

Free / trial

Free forever: 10,000 CPU seconds a month and 5 active workflows, no card required

Lowest paid plan

Pay as you go with no base fee: $0.00012 per CPU second beyond the free 10,000.

Commercial notes

The billing unit is the whole competitive argument. Per operation platforms charge for every node and action, so cost scales with workflow complexity as well as volume, and Latenode's own comparison puts 100,000 operations at roughly $19 on its metering against several hundred dollars on per operation platforms. Independent verification broadly supports the direction while producing different absolute figures, which is expected when the unit is time rather than count.

Key ambiguities

Third party verification puts the cheapest qualifying plan at $0 for 1,000 runs and $10.8 at 100,000 runs on three step workflows, while Latenode's own comparison cites roughly $19 at 100,000 operations. Both are plausible under different workflow shapes, which is itself the point: the unit is compute time, not runs.

Missing data

Enterprise rates and self hosted licensing terms.

Agentic Index verified 2026-09-29

Official pricing page

Related vendors

Alternatives to Latenode

The closest documented capability profiles to Latenode among agent builders tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

Sema4.ai13.0 / 14Fuller documented coverage on Testing, Debugging & Optimization
StackAI13.5 / 14Fuller documented coverage on Human Oversight & Guardrails and Testing, Debugging & Optimization
Airtable11.0 / 14A lighter documented profile than Latenode
Base4413.0 / 14Fuller documented coverage on Human Oversight & Guardrails and Memory & State Persistence
Cassidy12.0 / 14Fuller documented coverage on Human Oversight & Guardrails
Dify12.0 / 14Fuller documented coverage on Human Oversight & Guardrails

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Compare Latenode with other vendors →Compare all agent builders →See where Latenode ranks →

Data confidence: high