The best MTurk alternatives in 2026

Prices and facts checked on each company's own website and in AWS documentation on 24 October 2026.

Amazon Mechanical Turk closed permanently on 30 September 2026, and no single service replaces all of it. The right alternative depends on what you ran there. For ongoing business and AI data work, such as data entry, document extraction, moderation, transcription and AI output review, a managed team like Deepen AI Services does the work and checks it for you. Research panels such as Prolific and CloudResearch Connect supply participants for surveys and studies, and crowd marketplaces such as clickworker and Microworkers let you post and check microtasks yourself.

The short answer: match the alternative to the work

The alternatives compared

The alternatives in detail

1. Deepen AI Services: a trained team for ongoing data and AI work

Deepen AI Services replaces the anonymous crowd with a named, trained team. You send your HIT instructions and examples. The team is trained on them, works inside your tools or your Ground Truth and A2I workforce, and a team lead checks the work before it reaches you. Deepen AI has run trained data teams since 2017 and has more than 200 programs in production.

• Best for: Ongoing tasks with clear rules: data entry, document extraction, text labeling, product data, content moderation, transcription and AI output review.
• How you start: Describe the task or paste your HIT instructions. You get a plan and a price by email within 2 working days, with no sales call, then a two-week paid pilot on your real work.
• Pricing: $5 per person-hour for human review and data operations in English, with the team lead, QA sampling and a weekly report included. AI evaluation, robotics and other specialist work is quoted in your plan.
• Quality: Known answers, QA sampling and escalation are run by the team lead, and you see the results in a weekly report. Unclear items go to the lead, or to you, instead of each worker guessing.
• Security: SOC 2 Type II, ISO 27001, ISO 9001 and TISAX. In-house teams only, never crowdsourced, working on accounts you control, under NDA.
• Track record: Deepen AI has delivered annotation and data services for Daimler Trucks, BMW, Bosch, Mercedes, Hexagon, Aptiv and Ford Otosan.

See how a move from Mechanical Turk works, or get a plan for your tasks.

2. Prolific: participants for research and AI evaluation

Prolific is the alternative most often recommended to researchers who used MTurk to find participants. It lists more than 300,000 active participants in over 40 countries, with the strongest coverage in the US, the UK and Europe.

• Best for: Surveys, behavioral studies and evaluations that need screened, representative participants.
• How you start: Self-serve. Add funds to a wallet and launch a study, with no contract or subscription. Managed services are quoted separately.
• Pricing: The reward you set plus a platform fee: usually 42.8% for companies and 33.3% for academic and non-profit researchers. Prolific recommends paying at least $12 an hour and sets a minimum of $8.
• Good to know: You design the study and review the responses yourself.

3. CloudResearch Connect: a US research panel

CloudResearch built tools for researchers on MTurk for years, and Connect is its own participant platform. Participants are in the US and 18 or older.

• Best for: Academic and behavioral studies with US participants, especially teams that used CloudResearch's MTurk tools.
• Pricing: A 40% fee on participant pay for non-academic accounts and 25% for academic accounts. The recommended minimum pay is $7.50 an hour.
• Good to know: Like Prolific, you run the study and review the data.

4. clickworker: a crowd for standard tasks

clickworker, now part of LXT, offers a self-service marketplace for standard tasks and a managed service for custom projects. It describes its crowd as millions of people in 136 countries.

• Best for: Standard tasks at volume: tagging, categorization, surveys, store checks and web research.
• Pricing: On the marketplace, you pay per completed task. Managed projects are quoted by its sales team.
• Good to know: On the marketplace, you write the task and check the results.

5. Microworkers: do-it-yourself microtask campaigns

Microworkers is the closest to MTurk's self-serve model. You build a campaign from a template, add verification questions and review the submissions. Its homepage cites more than 4.9 million workers worldwide.

• Best for: Small, simple tasks you can design and verify yourself.
• Good to know: Fees aren't listed on its homepage, so check them in your account before you plan a budget.

6. Toloka: training and evaluation data for AI

Toloka now focuses on training and evaluation data for AI agents and LLMs: agent demonstrations, preference data, red-teaming and expert work. Its site points businesses to a conversation with its experts and shows no prices.

7. Appen and Scale AI: enterprise AI data programs

Both run large, sales-led programs for AI training data. Appen cites coverage of more than 500 locales, and Scale AI asks you to book an intro call to get started. Neither publishes prices. They're built for large, long-running AI data programs rather than the short batches many teams ran on MTurk.

8. CloudFactory: managed teams on annual commitments

CloudFactory offers managed teams for AI data. It starts with a free analysis of a sample of your data, then sends a proposal. There's no public rate card: pricing is built on annual spend commitments with custom rates.

9. Ground Truth or A2I, with a new workforce

If your pipeline sent work to MTurk through SageMaker Ground Truth or Amazon A2I, only the MTurk worker type is gone. Existing AWS customers can keep both services and point their jobs at a private workforce, such as your staff or a vendor's team, or at a vendor from AWS Marketplace. AWS says both services are no longer open to new customers and won't get new features.

The Ground Truth and A2I migration guide covers each step. Deepen AI's team can join your private workforce, so your task templates, S3 data and code stay as they are.

What each option really costs

Platform fees are only part of the price. On MTurk you paid a 20% fee on rewards, 40% for tasks with 10 or more assignments, plus your own time writing qualification tests and rejecting poor work. Compare alternatives on the cost of each accepted result, not the fee alone.

An example: if one person copies 120 receipts an hour, $5 an hour works out to about 4 cents a receipt, with the checking already done. On a self-serve platform, add the hours your own team spends reviewing and re-running rejected work.

How to choose in five questions

1. Is it a study with participants, or a task with right answers? Participants come from a research panel. Tasks with right answers, and the data work around a study, suit a trained team.
2. Is the work ongoing? Weekly work gets better with the same trained people. A one-off batch suits a marketplace.
3. Who will check the results? On self-serve platforms, you do. With a managed team, review is part of the service.
4. Where does the data live? If it's in Ground Truth, A2I, an admin panel or a spreadsheet, choose an option that works there without moving it.
5. What security do you need? Customer records, documents and AI outputs often need in-house teams, an NDA and audited controls such as SOC 2 and ISO 27001.

Moving off MTurk: a one-week plan

• Approve or reject every submitted HIT and pay bonuses before 30 October 2026. Untouched HITs are approved automatically
• Save your HIT instructions, qualification tests and examples of accepted and rejected work
• Download your results, and your transaction history before 28 January 2027
• Search your code for public-crowd/default to find Ground Truth and A2I jobs that used the MTurk workforce
• List each task with its weekly volume and deadline
• Choose an alternative for each task, using the five questions above
• Run a test batch against known answers before you move full volume

For every date and deadline, see the MTurk shutdown FAQ.