Description
Amazon Mechanical Turk (MTurk) is a crowdsourcing marketplace connecting businesses and developers, called Requesters, with a distributed workforce of individuals, called Workers, who complete tasks online. It's designed for work that's still easier for humans than machines, such as content moderation, data deduplication, survey participation, and image or data labeling for machine learning.
Read More
Requesters post Human Intelligence Tasks (HITs), set the reward they'll pay per task, and access workers either through a web interface or a direct API integration. MTurk is commonly used across three broad areas: building, managing, and evaluating machine learning workflows (for example, collecting and annotating training data, or running human-in-the-loop validation of model outputs); business process outsourcing, where a large task is broken into smaller microtasks distributed across many workers; and general human insights, such as surveys and user research.
Pricing follows a pay-per-task model: Requesters pay workers a reward they choose, plus a platform fee of 20% of that reward and any bonus paid. An additional 20% fee applies to HITs with ten or more assignments. There's a minimum fee of $0.01 per assignment or bonus payment. Optional add-ons include the Masters Qualification, which carries a further 5% fee, for accessing workers who've demonstrated high accuracy on specific task categories, and Premium Qualifications, with per-assignment fees starting at $0.05 depending on the qualification.
Amazon has announced that Mechanical Turk will permanently close on September 30, 2026. The company states it "regularly evaluate[s] our programs, tools, and services and make[s] adjustments based on those assessments," and following that review has decided to close the service. Current Workers and Requesters are directed to an FAQ page with guidance on preparing for the shutdown. The service remains live and operational in the meantime, with Requesters and Workers still able to sign in and use it.
Given the announced closure, MTurk is not a viable option for new, ongoing projects; existing users should plan for a transition to alternative crowdsourcing or data-labeling providers.
- CategoryWorkflow Automation
- Pricing ModelPay-as-you-go
- DeploymentCloud
- PlatformsWeb
Amazon Mechanical Turk's Core Features
Post Human Intelligence Tasks (HITs) to a global, on-demand workforce
Pay-per-task pricing model set by the Requester
Web interface and direct API access for task management
Support for machine learning data labeling and human-in-the-loop validation
Business process outsourcing through microtask distribution
Optional Masters Qualification for higher-accuracy workers
Optional Premium Qualifications for specialized worker pools
24x7 access to a distributed international workforce
How to use Amazon Mechanical Turk?
Sign in as a Requester: create or access a Requester account
Design a HIT: define the task, instructions, and reward amount
Set qualifications (optional): add Masters or Premium Qualifications to target specific workers
Publish the task: post the HIT for Workers to accept and complete
Review submissions: approve or reject completed work and issue payment
Integrate via API (optional): connect MTurk programmatically for larger-scale or automated workflows
Amazon Mechanical Turk's Use Cases
- ML dataset labeling
- Content moderation
- Survey research
- Business process microtasks
