How much does Pipedream cost in 2026?

October 2026 update: Workflows shutdown and Connect-only pricing

As of October 2026, Pipedream's documentation states that Pipedream Workflows and String will shut down on 31 March 2027. Workflows no longer accepts signups or subscriptions, existing workflows keep running until that date, and Workflows data will be deleted by 30 April 2027. Pipedream Connect is not affected. The Free, Basic and Advanced rows in the June 2026 table below describe Workflows allowances that are no longer offered to new users.

How Workflows credits worked (as of June 2026)

Pipedream is a developer-first integration platform, and unlike task- or operation-based tools it meters compute. As of June 2026, one credit equals 30 seconds of execution at 256 MB of memory, and credit cost scales with memory: a step configured at 1 GB consumes credits roughly four times as fast as the same step at 256 MB.

Former pricing tiers (as of June 2026)

The credit definition is the part to internalise, because it ties cost to how long a workflow runs, not how many steps it has.

Diagram source (Mermaid)

flowchart LR
  A[Workflow step runs] --> B[Execution time x memory]
  B --> C{1 credit = 30s at 256 MB}
  C --> D[512 MB doubles the rate]
  C --> E[1 GB quadruples the rate]

The cost trap: idle time still bills

Because credits track wall-clock compute, a step that sits waiting on a slow upstream API still burns credits the whole time it blocks. A workflow that calls a sluggish third-party endpoint synchronously, run a few thousand times a day, can consume most of a plan's daily credits doing nothing but waiting.

How Pipedream compares

Pipedream's compute model is unusual in the no-code category and is closest to a hosted serverless runtime with an integration layer on top. For fast API work it stretched a long way on the former Workflows free and Basic tiers; for non-technical operators a task-based tool like Zapier is simpler, and for teams that must keep execution in their own infrastructure, self-hosted n8n is the usual alternative.

Editor's Note: Across ShadowGen Pipedream builds in 2024-26, the credit model is honest but easy to misjudge. On one client integration a single 9-second upstream API call, run a few thousand times a day, quietly consumed most of a Basic plan's daily credits while the workflow sat idle waiting on the response. The fix that works: move slow or unreliable upstream calls behind a queue or a separate async workflow rather than blocking inside the main run. — Rafal Fila, ShadowGen