Skip to content
Agents tracked: 502 Releases (7d): 613 Agent status: 4 with issuesSince last limit reset: Claude 7d · Codex 5d For AI agents: Markdown Updated Nov 18, 2026

Temporal Cloud

by Temporal

Long processes that survive crashes.

Integrations and automation

WebsitePricingGitHub

1. Price

“Pay-As-You-Go Starting at $50 per million actions Get started for free Free $150 credits for 90 days”

Free tier: $150 credits for 90 days, no base monthly fee. As reported on the official pricing page, checked Oct 13, 2026.

temporal.io/pricing

Checked against the maker's official pages, pricing on Oct 13, 2026. Tap a verdict to see its source.How we checkedReport an error

What Temporal Cloud does

Temporal lets your app run long, multi-step processes, such as an order, a sign-up or an AI assistant's task, that carry on where they stopped after a crash or restart. Your own code runs on your machines, and Temporal Cloud is the hosted service that keeps track of each step. Temporal itself is open source and can also be run on your own servers.

  • Temporal Cloud is one of 17 APIs in the Integrations and automation category on AgentGid.
  • At $0.05 per 1,000 requests it is the 2nd cheapest of the 4 that publish a per-request price (cheapest: Trigger.dev, $0.025).
  • Its cheapest paid plan ($50 a month) is the 11th cheapest of the 15 here with a monthly plan.
  • It is one of 16 that can be started for free.
  • 6 of them ship an official MCP server; Temporal Cloud does not.
  • Its official SDK is the 2nd most downloaded in the category (11.5M a week).

What people build with it

Payments and orders that never half-finish

Charge the card, reserve stock and book delivery as one process; if a step fails, it is retried or the earlier steps are undone.

Processes that wait days for a person

An approval, a document review or a booking that pauses until someone clicks, then carries on, even if your servers restarted meanwhile.

AI assistants that resume after a crash

Each model call and tool call is a saved step, so a long agent task picks up the conversation instead of starting over and paying twice.

Data and video processing pipelines

Split a big job, such as processing thousands of files, into many steps that run in parallel and resume without redoing finished work.

Scheduled jobs and maintenance

Nightly reports, cleanups or account checks run on a schedule, with each run visible and retried when it fails.

Welcome emails spread over weeks Idea

Send a series of onboarding messages days apart for each new user, with no cron table to maintain and no lost steps after a restart.

Temporal Cloud popularity

11.3M
SDK downloads a week
+19.9% on the 4-week average
43
public agent repos use its SDK
1,252
GitHub stars, temporalio/sdk-python
SDK downloads per weekpypi temporalio 11,320,701

SDK downloads from npm and PyPI, weekly. Numbers as of Nov 17, 2026.

Pricing

Free tier

$150 credits for 90 days, no base monthly fee

Cheapest plan

$50 a month

Usage price

$50 per million actions

Plans

Added automatically: its SDK was in 16 of the public agent repos we sampled when it was added. Prices read from the official pricing page.

What one action costs

ActionPrice
1 million actions (pay-as-you-go, first 5M)$50
A test workflow run of 10,000 actions, including 10% support$0.55
1 GB-hour of active storage (open workflows)$0.042
Price history
  • A year of the pricing page in the Internet Archive (8 months): the lowest monthly figure went from $100 (2025-12) to $100 (2026-09).

As reported on the official pricing page, checked Oct 13, 2026. Prices change: confirm before you buy. temporal.io/pricing

Limits

All Temporal Cloud namespaces (On-Demand)

500 actions per second by default, scaling up with the last 7 days of use

All Temporal Cloud namespaces

2 MB per request payload

All Temporal Cloud namespaces

Finished workflow history kept 30 days by default, settable from 1 to 90 days

From the official docs. docs.temporal.io/evaluate/cloud/limits.md

Key and access

How it signs in

API key (or mTLS certificate) per Namespace, passed to Client.connect with tls=True

Environment variable
TEMPORAL_API_KEY (with TEMPORAL_ADDRESS and TEMPORAL_NAMESPACE)
Safe in browser code
No, keep it on your server
Card at sign-up

Yes

Steps only a person can do

  • Sign up for Temporal Cloud with a credit card (or through AWS or GCP Marketplace) and create a Namespace. Use the Namespace address and ID from the Namespaces tab as TEMPORAL_ADDRESS and TEMPORAL_NAMESPACE.
  • Create an API key and save the secret right away; it is shown only once. Put it in .env as TEMPORAL_API_KEY (kept out of git) and connect with tls=True.
  • Install the Temporal CLI for local development (brew install temporal on macOS). Run temporal server start-dev and test workflows locally before using the cloud.

Commercial use and your data

Temporal Cloud is a paid service under Temporal's terms, and you keep the rights to the data you send. Temporal itself is open source under the MIT license, free to run on your own servers.

In a paid app
With conditions

Temporal Cloud is used under Temporal's terms on a paid account; the open-source server and SDKs are MIT and free for commercial use.

“Customer shall retain all right, title and interest in and to the Customer Data, including all intellectual property rights therein.” source

redistribute-files
With conditions

The MIT code may be shipped with the notice kept; access to the hosted service itself may not be resold.

“The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.” source

client-work
With conditions

Fine with the open-source version; on Temporal Cloud, the client's account and data stay under the client's own contract.

“Customer is solely responsible for the selection of the appropriate processing region based on localization requirements.” source

Trains on what you send

“Data conversion runs on your Workers and Clients, so Temporal Cloud cannot see or decrypt your data.”

Keeps your data

“Temporal Cloud sets the default Retention Period to 30 days.”

Where data is stored

“Create Namespaces in a region close to where you host your Workers to reduce latency.”

Errors to handle

ResourceExhausted

You went over the namespace's actions or requests per second and Temporal Cloud throttled the call.

The SDK retries it; if it keeps failing, handle the failure, slow down or raise capacity.

Request unauthorized

The API key is wrong, or a new key has not reached the namespace yet.

Retry with backoff for a few minutes; then check the key, its permissions and the namespace.

WORKFLOW_TASK_FAILED_CAUSE_PAYLOADS_TOO_LARGE

An input or result is bigger than the 2 MB payload limit.

Store large data elsewhere (for example in object storage) and pass a reference; deploy the fix and the workflow continues.

Context: deadline exceeded

A request to the Temporal service could not finish in time (network, overload or clock skew).

Retry; check the network and worker clocks; on Temporal Cloud open a support ticket with the workflow IDs.

NondeterminismError

Workflow code did something that differs between runs, such as calling the network or reading the clock directly.

Move that work into an Activity and redeploy; the workflow waits until it is fixed.

Use it from an agent

Our agent check

3 of 6: llms.txt yes, OpenAPI no, SDK yes, MCP no, status page no, AI assistants allowed yes

Checked Nov 13, 2026.

MCP server

Temporal Docs MCP Server (documentation search only)

Sign-in is required through MCP OAuth with Google or GitHub, and anonymous requests are rejected.

PyPI
pip install temporalio

Instructions in its docs to check first

The docs say

“For questions about Temporal (Workflows, Activities, Workers, SDKs, Temporal Cloud, the CLI, self-hosting), you can query the documentation directly instead of relying on web search or training data.”

The page asks agents to send the user's questions and a one-sentence job-to-be-done to a Temporal search endpoint. Ask the person first, and never send project details; reading the .md pages is enough.

The docs say

“If a page is wrong, out of date, or missing something you needed, or Temporal behaves differently from what the docs say, you can report it to the Temporal docs team.”

Sending a report is a message on the person's behalf; do it only if they ask.

Alternatives

Temporal Cloud This page

$0.05 / 1k requestsFree tier

Inngest

Pick Inngest for event-driven background functions with durable steps, at a similar per-step list price ($0.05 per 1,000 in our catalog on 2026-11-10).

$0.05 / 1k requestsFree tier

Trigger.dev

Pick Trigger.dev for long-running TypeScript tasks on a managed platform that also runs your code, listed at $0.025 per 1,000 in our catalog on 2026-11-10.

$0.025 / 1k requestsFree tier

Prefect Cloud

Pick Prefect Cloud when your work is mostly scheduled data pipelines in Python that you want to schedule and monitor.

Free planFree tier

All integrations and automation APIs

Get started

Tell your AI assistant

“Build my order process as a Temporal workflow with temporalio at a fixed version: each outside call (payment, email, shipping) as an Activity with a retry policy, test it on temporal server start-dev, then read TEMPORAL_ADDRESS, TEMPORAL_NAMESPACE and TEMPORAL_API_KEY from .env for Temporal Cloud.”

  1. Install the Python package in a virtual environment (there are also Go, Java, TypeScript, .NET, PHP, Ruby and Rust packages).

    pip install temporalio

    From docs.temporal.io/develop/python/set-up-your-local-python.md

  2. Start the free local development server; it runs the web screen at localhost:8233 and stores data only in memory.

    temporal server start-dev

    From docs.temporal.io/develop/python/set-up-your-local-python.md

  3. For Temporal Cloud, sign up, create a Namespace with an API key, and connect with these environment variables.

    export TEMPORAL_ADDRESS=<namespace>.<account>.tmprl.cloud:7233
    export TEMPORAL_NAMESPACE=<namespace>.<account>
    export TEMPORAL_API_KEY=<key-secret>

    From docs.temporal.io/cloud/get-started.md

First request

starter.py from the Python quickstart (blank lines removed), against the free local dev server (temporal server start-dev) with worker.py running. For Temporal Cloud the same code connects with Client.connect(endpoint, namespace=..., api_key=..., tls=True).

import asyncio
import uuid
from temporalio.client import Client
async def main():
    client = await Client.connect("localhost:7233")
    result = await client.execute_workflow(
        "SayHelloWorkflow",
        "Temporal",
        id=f"say-hello-workflow-{uuid.uuid4()}",
        task_queue="my-task-queue",
    )
    print("Workflow result:", result)
if __name__ == "__main__":
    asyncio.run(main())

Mistakes to avoid

Calling an outside service, reading the clock or using random values directly inside workflow code.
Put outside calls in Activities and use the workflow module's own helpers inside workflows.

Workflow code is replayed, so anything that changes between runs breaks it and the workflow stops until fixed.

Importing activity modules into the workflow file normally.
Wrap those imports in with workflow.unsafe.imports_passed_through(), as the quickstart does.

Workflows run in a sandbox that re-imports the workflow file on each run.

Passing large files or long texts as workflow inputs or results.
Store the data elsewhere and pass an ID or link.

Each request payload is limited to 2 MB on Temporal Cloud.

Starting workers right after creating a new API key.
First check the key with a simple call (for example temporal workflow list) and retry with backoff, then start workers.

New keys reach the namespace asynchronously and can return Request unauthorized at first.

Letting an Activity retry forever on an error that will never fix itself, such as bad input.
Raise ApplicationError with non_retryable=True for permanent errors.

Otherwise the step keeps retrying until its timeout.

Questions

Is there a free tier?

There is no free plan, but new accounts get $150 in credits for 90 days and no base monthly fee. Running Temporal locally or on your own servers is free.

Does Temporal Cloud see my data?

It stores the inputs and results of each step. You can encrypt them on your side first, and then Temporal Cloud cannot read them.

Which languages can I write workflows in?

Go, Java, Python, TypeScript, .NET, PHP, Ruby and Rust, as listed on the pricing page.

What happens when a step fails?

Activities are retried automatically with backoff; you can mark errors that should not be retried so the step fails at once.

Can my AI assistant read the docs?

Yes. Any docs page is available as Markdown by adding .md to its address, with an index at docs.temporal.io/llms.txt.