Trigger-and-action automations on data in Airtable bases, from a single notification to multi-step workflows. · Hay que volver a comprobar la evidencia
Precio documentado24 US$/mes por usuarioHay que volver a comprobar el precioCondiciones de precio
Precio documentado: 24 US$/mes por usuario · Plan: Team · permanencia de 1 mes
Temporal describes a durable execution platform for developers.
Precio documentado50,00 por 1,000,000 Actions at the starting rate (excedente) (moneda no indicada)Condiciones de precio
Precio documentado: 50,00 por 1,000,000 Actions at the starting rate (excedente) (moneda no indicada) · Plan: Pay-As-You-Go · permanencia no confirmada
Precio documentado: 50,00 por 1,000,000 Actions at the starting rate (excedente) (moneda no indicada) · Plan: Pay-As-You-Go · permanencia no confirmada
Active scenarios: 2 on Free, unlimited on paid plans; Minimum interval between scheduled runs: 15 minutes on Free, 1 minute on paid plans; Maximum run time: 5 minutes on Free, 40 minutes on paid plans
Plan-table columns are Free, Core, Pro, Teams and Enterprise.
Pipedream Workflows and String shut down on March 31, 2027; Workflows are in maintenance mode: existing workflows run until then, with no new features; New Workflows sign-ups and subscriptions are no longer accepted; Pipedream Connect is not affected
All Workflows data, including credentials and OAuth tokens, is deleted by April 30, 2027. An export tool downloads workflow structure as JSON, one project at a time. Subscriptions are cancelled on March 31, 2027 with a prorated refund.
Users: up to 5 on Free and Plus, 25 on Team, unlimited on Ultimate; Credits: 1,000 on Free, 10,000 a month on Plus, 50,000 a month on Team; Over the allowance, Free runs wait for tomorrow; Plus and Team pay USD 0.007 per credit
Team adds unlimited projects, SSO and email support.
Concurrency: 1 job at a time on Standard and Business MCP, 15 on Enterprise MCP, 30 on Workato ONE
Concurrency is the number of recipe jobs processed at the same time; quantities are totals per edition. The passage starts at the end of the Standard block.
Trigger steps, Filter and Paths steps, failed or halted actions and skipped steps use no tasks; Tables, Forms, Formatter, Delay, Looping, Sub-Zap, Digest, Zapier Manager and Storage steps use no tasks; Zapier Agents use a separate activity quota, not tasks; Example: a Zap with two actions that runs ten times uses 20 tasks
A task is any successful action. Some steps cost more: each lead routed by Lead Router uses 5 tasks and each Zapier MCP tool call uses 2. Zaps, AI steps, code, MCP and the SDK share one task allowance per account.
Each module action, such as reading a record, writing a spreadsheet row or posting a message, uses one credit; Reading, searching, creating, updating and deleting data, transforming it, aggregating rows and iterating rows all count; One scenario run can use from two to thousands of credits
Router and error-handler modules (Rollback, Break, Resume, Commit, Ignore) use no credits. Make Code App steps use 2 credits per second of execution time.
External user: each distinct user of your app who connects at least one account; one user with five connected apps counts once; Usage credit: 30 seconds of compute for an action run, an MCP tool call, a Connect proxy request or an event emitted by a deployed trigger; Listing apps and accounts, configuring components and other management calls use no credits
Workflows, until the shutdown, use one credit per 30 seconds of compute per workflow segment at 256 MB; step count does not matter, and doubling memory doubles credits.
Triggers do not count as Tasks, whatever the connector; Each action step counts as 1 Task, in any connector; Control-flow steps (IF, loops, Stop job, error handlers) do not count; steps inside them count each time they run, once per loop iteration; Successful steps count even if the job later fails; Document-processing pages and Event streams events are measured separately, not as Tasks; A bulk or batch action counts as 1 Task however many records it processes; AI by Workato action steps count as more than one Task each, at a rate set in your contract
Workato example: one action plus two actions in a loop over 5 rows uses 11 Tasks. Some contracts measure Business actions or Workato credits instead of Tasks.
Every step that runs is a task, including the webhook trigger; Loop and boolean steps count each time they run, once per list item, plus the loop's final check; Example: a webhook delivering 2 records through Clearbit, a loop, a boolean and Slack used 9 tasks in one run
Tray describes a task as roughly one workflow step each time it runs; a connector that makes several API calls in one step counts once. Retries are not charged. Each Connectivity API Call Connector request is 1 task.
Power Platform requests: every connector call, HTTP action and built-in action (even initialising a variable or a compose step) counts, successful or failed, including retries and pagination.
A flow is made of steps of two types: a trigger step starts the flow, and the action steps after it define what is done with the incoming data from previous steps.
Publisher key-concepts documentation (packages/docs/pages/guide/key-concepts.md) at release v0.15.0 (commit 7a1e5e2d, the latest GitHub release; the default branch has later commits) captured on 9 November 2026; supplier description, not a deployment or independent test. The page defines the unit of building, not a metered or billable unit; no usage price is stated for self-hosted Automatisch.
Airflow workflows are authored as Dags that orchestrate tasks; the scheduler executes the tasks on an array of workers while following the specified dependencies.
Apache Airflow repository README at release 3.4.0 (commit 619a787a) captured on 9 November 2026; supplier description, not a deployment or independent test. The README describes the unit of work, not a metered or billable unit; self-hosted Airflow has no usage price in the README.
Dagster+ credit usage is the sum of one credit per step execution and one credit per asset materialization event; each partition emits a separate materialization event.
Publisher documentation in the official dagster-io/dagster repository at release 1.13.29 (commit 7dca97fa), captured 9 November 2026; supplier description, not a deployment or independent test. Applies to the managed Dagster+ offering; the per-credit prices were not captured.
Actions are the billable operations between your application and Temporal Cloud; The page lists Workflow, Activity, Timer, Signal, Query and Schedule under Actions; Workflow event history is stored as Active storage for open workflows and Retained Storage for closed workflows; Storage starts at $0.042/GBhr active and $0.00105 retained
Temporal Cloud pricing page captured 9 November 2026; supplier statement, not a purchase or test. The pricing page does not list which operations count as an Action within each type; see its Actions documentation. The storage rates are shown without an ISO currency code.
Unused credits expire at the end of the term; Extra credits expire after one month on monthly subscriptions, or at the end of the year on annual Pro and Teams subscriptions
With yearly billing, prepaid credits last 12 months instead of expiring each month.
Usage draws on one purchased balance of Workato credits shared by workflows, document processing and agents.
No credit or Task allowance is published; some plans measure usage differently. Add-ons such as extra concurrency and on-premises agents are billed separately from credit usage.
Automations per base: 75 on Free, 100 on Team, 150 on Business, 200 on Enterprise Scale; Runs per month: 100 on Free, 25,000 on Team, 100,000 on Business, 500,000 on Enterprise Scale
Compare-table columns are Free, Team, Business and Enterprise Scale.
Dagster+ Serverless is limited to 100 GB of bandwidth and 4,500 step-minutes per day; runs receive 4 vCPU cores, 16 GB of RAM and 128 GB of ephemeral disk, and code locations 0.25 vCPU cores and 1 GB of RAM. Dagster+ Pro customers may request a quota increase.
Publisher documentation in the official dagster-io/dagster repository at release 1.13.29 (commit 7dca97fa), captured 9 November 2026; supplier description, not a deployment or independent test. Limits of the managed Dagster+ Serverless deployment, not of open-source Dagster you host yourself; no run was executed and plan prices were not captured.
An operation is any module run, including one that only checks for new data; A polling trigger such as Watch New Rows uses one operation per check, however many records it returns; Each later module runs once per returned record (bundle), so records multiply its operations
By default one operation is one credit (Make help on credits).
An event source from the public registry, including a polling source, runs free when it triggers a workflow, on every plan; A private custom source is charged for all compute time, including the first 30 seconds
Real-time triggers react to events from external systems, such as file arrivals and new messages on message buses (Kafka, Redis, Pulsar, AMQP, MQTT, NATS, AWS SQS, Google Pub/Sub, Azure Event Hubs). The README states no minimum schedule interval.
Dated publisher project README at release v2.0.9 (commit b961b225) captured 9 November 2026; supplier description, not a deployment or independent test. Describes the product design; no trigger was run and no latency was measured.
The Inject node can trigger a flow manually from the editor or automatically at regular intervals; the interval can be set up to a maximum of 596 hours (approximately 24 days), and for intervals greater than one day the page suggests a scheduler node that can cope with power outages and restarts.
Official Node-RED documentation page on nodered.org captured 9 November 2026; supplier description, not a hands-on test. Describes the Inject node; no flow was run.
The README says Huginn can scrape websites and send an email when they change, and watch RSS feeds, IMAP mailboxes, webhooks and MQTT topics. It states no polling intervals.
Dated publisher project README at release v2026.11.09 (commit 639c050b) captured 9 November 2026; supplier description, not a deployment or independent test. Describes the product design; no Agent was run.
Deployments can run on Cron, Interval or RRule schedules; the schedules page states no minimum interval.
Publisher documentation in the official PrefectHQ/prefect repository at release 3.9.2 (commit b0479c0c), captured 9 November 2026; supplier description, not a deployment or independent test. Interval schedules accept seconds, ISO 8601 durations or time format strings; no schedule was run.
A sensor's minimum_interval_seconds sets the minimum number of seconds between evaluations; it is a floor, not the exact frequency, and a sensor that takes longer than the interval delays the next evaluation.
Publisher documentation in the official dagster-io/dagster repository at release 1.13.29 (commit 7dca97fa), captured 9 November 2026; supplier description, not a deployment or independent test. Describes the sensor setting in the documentation; no sensor was run and no default interval was captured.
1,200+ pre-built connectors included for every user; Unlimited workspaces and users
Extra concurrency and on-premises agents are add-ons billed separately from usage; other listed add-ons include RPA, EDI, Private Link, Workflow Apps and Virtual Private Workato.
The README says BuildShip workflows can connect to third-party APIs and tools, integrate with AI models and run database workflows for Firestore, Supabase and Postgres.
Publisher repository README at the main branch (commit d4cf33af, last changed 11 June 2024; the repository has no release) captured on 9 November 2026; supplier description, not a hands-on test. The README is old and current service terms were not confirmed. Which integrations are included in which plan, and any extra charge for them, is not stated in the README.
Node-RED ships a core set of nodes; many more come from the Node-RED project and the wider community through the Node-RED library and can be installed from the editor's Palette Manager. The page describes no paid connector tier.
Official Node-RED documentation page on nodered.org captured 9 November 2026; supplier description, not a hands-on test. Individual community nodes have their own licences, quality and maintainers, which were not reviewed.
The README’s MODE table marks agent mode as (EE), It marks the indexer mode (full-text search) as needing the EE image
Dated publisher project README at release v1.865.0 (commit 2c1edd4c) captured 9 November 2026; supplier description, not a deployment or independent test. Taken from the README’s environment-variable table, where EE is the README’s abbreviation for the Enterprise Edition (its LICENSE_KEY row says Enterprise Edition); other Enterprise-only features are not listed in the README.
The README lists things you can do with Huginn: post to Slack, Telegram, Threads, Tumblr, Twitter, Weibo, Pushover, Pushbullet and Jabber, send SMS and phone calls with Twilio, and fetch, transform and store files on Dropbox, S3, FTP and the local filesystem. It does not mention a paid connector tier.
Dated publisher project README at release v2026.11.09 (commit 639c050b) captured 9 November 2026; supplier description, not a deployment or independent test. Whether each Agent still works with its third-party service was not tested; those services have their own terms and prices.
The documentation lists supported apps including GitHub, GitLab, Gmail, Google Calendar, Google Drive, Google Forms, Google Sheets and Google Tasks; the list does not mark any of them as paid.
Publisher documentation page Available Apps (packages/docs/pages/guide/available-apps.md) at release v0.15.0 (commit 7a1e5e2d, the latest GitHub release; the default branch has later commits) captured on 9 November 2026; supplier description, not a deployment or independent test. The same list includes HTTP Request and Webhooks; the page does not say whether any app is limited to the commercial Enterprise Edition.
The README says the power of Airflow also comes from a number of providers that extend the core functionality and are released separately from Airflow Core.
Apache Airflow repository README at release 3.4.0 (commit 619a787a) captured on 9 November 2026; supplier description, not a deployment or independent test. The README also says the framework includes a wide range of built-in operators and can be extended, and that provider versions follow their own SemVer numbers (for example google 4.1.0 and amazon 3.1.1). It does not state whether any provider is paid or restricted.
The README says BuildShip projects are created in a region of your choice, that your data and workflows stay in that region, and that each project gets a single-tenant GCP project dedicated to you.
Publisher repository README at the main branch (commit d4cf33af, last changed 11 June 2024; the repository has no release) captured on 9 November 2026; supplier description, not a hands-on test. The README is old and current service terms were not confirmed. It describes the hosted service, not a self-hosted deployment; the README says exporting and deploying workflows on your own cloud is a plan that was still work in progress.
The getting-started guide lists running Node-RED locally, on a Raspberry Pi, with Docker, from git, on BeagleBone boards, on Android (experimental, via Termux), with FlowFuse (multi-tenant), on AWS (Elastic Beanstalk or EC2) and on a Microsoft Azure virtual machine.
Official Node-RED documentation page on nodered.org captured 9 November 2026; supplier description, not a hands-on test. No installation was performed.
Runs on AWS (EKS/ECS), GCP, Azure, Ubicloud, Fly.io, Render.com, Hetzner, Digital Ocean and others; the README’s rule of thumb is one worker per vCPU and 1-2 GB RAM
Dated publisher project README at release v1.865.0 (commit 2c1edd4c) captured 9 November 2026; supplier description, not a deployment or independent test. The README also documents Docker Compose and Kubernetes (Helm) installation. Sizing is a supplier rule of thumb, not a measured requirement.
Huginn runs as a web process plus a background worker backed by MySQL or PostgreSQL. The README recommends Docker Compose on your own server for production, with official images for all processes in one container or one process per container.
Dated publisher project README at release v2026.11.09 (commit 639c050b) captured 9 November 2026; supplier description, not a deployment or independent test. No deployment was performed. The README also describes Dokku, managed container platforms, community packages for YunoHost and Unraid, and manual installation on Ubuntu or Debian.
Work pools come in three types: Hybrid (a worker in your infrastructure submits runs to your infrastructure), Push (runs are submitted automatically to your configured serverless provider) and Managed (runs are submitted automatically to Prefect-managed infrastructure).
Publisher documentation in the official PrefectHQ/prefect repository at release 3.9.2 (commit b0479c0c), captured 9 November 2026; supplier description, not a deployment or independent test. The Managed type uses Prefect-managed infrastructure and is listed by the documentation under Prefect Cloud, not under the self-hosted server.
Dagster+ Serverless is a fully managed version of Dagster+ in which you run Dagster jobs without spinning up any infrastructure yourself. The documentation says it suits workloads that mainly orchestrate other services or do light computation, and introduces conditions under which to select Hybrid deployment instead.
Publisher documentation in the official dagster-io/dagster repository at release 1.13.29 (commit 7dca97fa), captured 9 November 2026; supplier description, not a deployment or independent test. Describes the managed Dagster+ deployment types, not the open-source software; Dagster+ plan prices and per-plan availability were not captured.
AI by Zapier steps and their tool calls use 1 task on standard models or with your own model provider (you also pay the provider), 3 tasks on advanced models and 5 on premium models.
Tool calls are not available on the standard tier. Zapier MCP tool calls use 2 tasks each. Zapier says rates may change with notice.
Make's AI Provider is available on all plans in Make AI Agents and Make AI Toolkit; Credits are charged by input and output tokens and the model chosen; Small and Medium tiers: 18,080 input or 2,260 output tokens per credit; Large tier (GPT-5 mini): 3,616 input or 452 output tokens per credit
With your own OpenAI or Anthropic connection (paid plans), Make charges credits per operation and the provider bills the tokens.
Assistant credits for AI-assisted building: 1,600 a month on Starter, 4,000 or 9,600 on Pro, none on Enterprise; Free trials include 800 Assistant credits; Credits refresh monthly, do not roll over, and extra credits can be bought; On self-hosted n8n you bring your own AI key and no credit limit applies
Assistant is in preview. The price of extra credits is not stated.
AI steps add credits on top of the run: 2 on a fast model, 10 on a smart model, 20 on a frontier model; Agentic actions are 1 credit each; On Plus and higher, your own AI provider keys make every AI step 1 credit, with the model cost billed by your provider
Genies (agents) are measured per prompt sent, by users or by systems; Skills are included when a Genie invokes them and billed as Tasks when invoked over MCP
AI by Workato action steps count as more than one Task each, at a contract rate.
AI connectors convert input and output tokens into tasks at a token-to-task ratio set in your contract, available from support.
Merlin Text Generation, Merlin Text Analysis and AI-Agent-Utils generate token usage; document processing (Merlin IDP) counts pages. Agent Development (Agent Gateway for MCP and Merlin Agent Builder) is an add-on.
AI-powered runtime such as agents is outside the request limits and consumes Copilot Credits.
Agent flows are metered through Microsoft Copilot Studio, not Power Platform requests. Copilot Studio is sold as prepaid Copilot Credit Commit Units, up to 20% cheaper bought up front, with pay-as-you-go coverage when prepaid credits run out; no credit price is shown.
The README says Huginn can summarize, classify and translate text, or generate images and speech, with OpenAI-compatible APIs. It does not say who supplies or pays for the API access.
Dated publisher project README at release v2026.11.09 (commit 639c050b) captured 9 November 2026; supplier description, not a deployment or independent test. Charges of whichever OpenAI-compatible API is used are not described in the README and were not captured.
Tasks beyond the plan limit are charged separately through pay-per-task billing; Pay-per-task stops at three times the plan task limit; further runs are held until the billing cycle ends; Without pay-per-task, new Zap runs are held at the task limit until the billing cycle resets
Pay-per-task tasks cost more than plan tasks; the rate depends on plan and billing cycle and is shown only in account billing settings. It is on by default for accounts created after January 2024.
When credits run out, scenarios stop until credits are added; Incoming webhooks queue up to the plan webhook queue size; Extra credits come in bundles of 1,000 or 10,000 at a price set in your subscription; Core, Pro and Teams can auto-purchase 10,000 credits before running out
Make notifies at 75% and 90% of purchased credits. The bundle price is not published on the pricing page.
Plus and Team keep running past the allowance at USD 0.007 per extra credit; on Free, runs past the quota wait until credits refresh; Ultimate uses a yearly pool.
No per-request charge is listed; Microsoft may enforce limits by throttling; Capacity is added to the user or flow that exceeds its limit; Each Power Platform requests add-on raises a limit by 50,000 requests per 24 hours, and several can be assigned
A cloud flow that is throttled consistently for 14 days is turned off (Microsoft Learn limits page). The add-on price is not on the pricing page.
Current public website plan model captured 3 November 2026. Existing-account migration, app-store offers and country-specific availability are not established; confirm the applicable account and offer. Supplier documentation, not hands-on testing.
Publisher project documentation captured 3 November 2026; a description of the selected product and source version, not a hands-on test. Current service availability, pricing and terms require confirmation.
Publisher project documentation captured 3 November 2026; a description of the selected product and source version, not a hands-on test. Current service availability, pricing and terms require confirmation.
Publisher project documentation captured 3 November 2026; description of the selected source version, not a hands-on test. Current availability, pricing and terms require confirmation.
Publisher project documentation captured 3 November 2026; description of the selected source version, not a hands-on test. Current availability, pricing and terms require confirmation.
Publisher repository README at the captured revision, 4 November 2026. No installation, external service, action, account or automation was tested. Third-party service permissions and costs are separate.
Publisher repository README at the captured revision, 4 November 2026. No installation, external service, action, account or automation was tested. Third-party service permissions and costs are separate.
Project repository README at the captured revision, 4 November 2026. No installation, account, workload, integration or performance test. Exact release, hosted services and third-party terms require separate review.
Project repository README at the captured revision, 4 November 2026. No installation, account, workload, integration or performance test. Exact release, hosted services and third-party terms require separate review.
Project repository README at the captured revision, 4 November 2026. No installation, account, workflow, integration or performance test. Exact release, hosted services and third-party terms require separate review.
Project repository README at the captured revision, 4 November 2026. No installation, account, workflow, integration or performance test. Exact release, hosted services and third-party terms require separate review.
Cara a cara
Comparaciones destacadas de automatización de flujos
¿Qué herramientas de automatización de flujos tienen plan gratuito?
Zapier, Make, n8n, Pipedream, Activepieces, IFTTT, Airtable, Kestra, Node-RED, Windmill, Huginn, Automatisch, Apache Airflow, Prefect, Dagster y Temporal documentan un plan gratuito en sus páginas oficiales de precios. Los planes gratuitos suelen limitar el uso, los usuarios o las funciones; revisa los límites en cada perfil.
¿Cuánto cuesta el software de automatización de flujos?
Entre los precios de entrada documentados de los planes de pago están Zapier desde 29,99 US$/mes, Make desde 9 US$, n8n desde 24 €/mes, Pipedream desde 150 US$/mes y Activepieces desde 20 US$/mes. Workato y Tray.ai solo trabajan con presupuesto. Los precios proceden de las páginas oficiales en las fechas indicadas; las unidades y los periodos de facturación varían, así que compara opciones equivalentes.