Apify Server Actors are Actors that stay ready and answer requests like an API, instead of starting a fresh run for every job. Apify announced general availability on 10 November 2026. They fit short, frequent requests from apps and AI agents. A normal scheduled run is still the better fit for batch scraping, daily monitoring and anything that runs once and stops.

What did Apify announce on 10 November 2026?

According to Apify's press release, Server Actors are now generally available in Apify Store. The same release says weekly Actor runs passed 267 million as of October 2026, more than five times the level at the end of 2025, and that the Store passed 100,000 Actors built by more than 4,500 independent developers and official partners.

Apify's documentation describes Server Actors as the feature formerly called Actor Standby mode. An application or an AI agent sends a request, gets a response, and does not wait for a new container to start each time.

How is a Server Actor different from a normal Actor run?

QuestionNormal runServer Actor
How it startsYou start a run with an inputIt stays ready and takes requests
Typical callerSchedule, n8n, a script, a personAn app or an AI agent
ResultA dataset you read after the runA response to each request
Idle timeNone, the run endsResources are used while it waits, until the idle timeout stops it (per Apify docs)

The last row matters. Apify's docs say a run also consumes resources while it waits for requests, so you should expect usage even when no requests arrive. Apify's press release does not give Server Actor pricing, so check the Actor's own pricing tab before you build on one.

When should you still use a normal Actor run?

Most scraping jobs are not request and response jobs. Use a normal run when:

  • The job is a batch. Scraping 5,000 listings, a full directory or a whole catalogue is one long run that ends in a dataset.
  • The job is on a schedule. A daily price check or a weekly report does not need something waiting around between runs.
  • You want to pay only for work done. A run that ends costs nothing while it is not running.
  • You want a clean record. Each run gives you a dataset, a log and a start and end time you can audit.

For example, Website Change Monitor is built for a scheduled run that compares a page with its last version and alerts you. Waiting for requests would add nothing. The same goes for Dataset Diff, which keeps only new and changed items between two runs of any Actor.

When does a Server Actor make sense?

Think of cases where a person or an agent asks a small question and needs the answer quickly:

  • An AI agent that checks one brand, one page or one product during a conversation.
  • An internal tool where a salesperson types a domain and wants a quick answer.
  • A workflow that calls the same Actor many times a day with small inputs.

Our own tools are mostly batch tools, so the honest summary is this: if you can describe the job as "run it, then read the dataset", you do not need a Server Actor. If you can describe it as "ask it a question and wait for the answer", look at one.

How do AI agents reach Actors today?

There are two directions, and they are easy to mix up. The Apify MCP server exposes Actors as tools to an outside AI client. MCP connectors, launched by Apify on 9 June 2026, go the other way and let an Actor call third-party services such as Notion, Slack, GitHub, Sentry and Supabase through MCP, using credentials you set up once in the Console. Server Actors are a third piece: they change how an Actor is hosted, not who calls whom. Our guide to the Apify MCP server for AI agents covers the first direction in more detail.

If your agent only needs a few well-known pages watched, an example is the AI Brand Visibility Tracker, which checks how ChatGPT, Perplexity and Gemini mention your brand. That is a scheduled check, so a normal run is the right fit.

Is there another Apify change you should check this week?

Yes, if you read the Store programmatically. Apify's changelog says that on 15 November 2026 GET /v2/store moves to a new search backend. Per that entry, the total and count fields now report the number of items returned rather than the number of matches, so a loop that runs until the offset reaches the total will stop after one request. The sortBy parameter is still accepted but has no effect: searched results come back by relevance and unsearched listings follow Store order. The change also applies to apify-client and the Apify MCP server. If you need a particular order, sort the results yourself. We did not test the new backend, so confirm the behaviour against your own code.

A quick checklist before you choose

  1. Does the job end by itself? Use a normal run.
  2. Does it run on a schedule? Use a normal run and a schedule.
  3. Does something call it many times a day with small inputs and wait for the answer? Look at a Server Actor and read its pricing and idle timeout first.
  4. Do you need the output in a sheet or an alert? Add a workflow, for example in n8n, after the run. See our n8n 3.0 checklist for scraping workflows.

For the full list of tools by job, browse the developer tools hub.

FAQ

What is an Apify Server Actor?

An Actor that stays ready and handles requests like an API, so apps and AI agents get a response without starting a new run each time. Apify's docs say it is the feature formerly called Actor Standby mode.

Are Server Actors free to run while idle?

Apify's docs say a run also consumes resources while it waits for requests, until the idle timeout stops it. Expect usage even when no requests arrive, and check the Actor's pricing before you rely on it.

Do I need a Server Actor to scrape on a schedule?

No. A scheduled normal run ends when the job ends and leaves a dataset, which suits daily or weekly scraping and monitoring.

What is the difference between the Apify MCP server and MCP connectors?

The MCP server exposes Actors as tools to outside AI clients. MCP connectors let Actors call external MCP servers as tools. They run in opposite directions.

Try it

You can start with a free Apify account and test any of our Actors on a small input first: see all our Actors on Apify.

Last updated: November 2026

Need help implementing this?

I build custom automation, scraping pipelines, and AI solutions for businesses. 170+ projects delivered with a 5.0 rating across 174 reviews. Tell me about your project - I reply within 24 hours.

Start Your Project →