Mercor Review: Pay, AI Interview and What the Work Is Really Like

Every Mercor review ends up at the same number, so let us open with it: $200 an hour. That is the top of the range credentialed specialists report for evenings spent grading what AI models write. A cardiologist corrects what a model says about heart failure. A tax lawyer marks up its advice on a filing. Both bill it from their own kitchen table.

The platform sits between the biggest AI labs and the experts who teach their models. You upload a profile, sit a short AI interview, and Mercor matches you to contracts that pay by the hour in US dollars. Credentialed specialists are exactly who the labs lack, so that is where the money concentrates.

Below: the rates by specialty, how the AI interview screens you, how quickly the money lands, what improves your odds of a match, and how Mercor compares with Outlier and DataAnnotation.

Mercor is a talent marketplace built for one specific buyer: AI labs that need human experts to train and grade their models. Instead of posting a job and reading thousands of CVs, a lab tells Mercor what kind of expert it needs. Mercor then screens candidates with an AI interview and matches the ones who fit.

What does Mercor do for you, then? It handles the finding, the vetting, the contract and the payment. You supply the expertise. The work stays remote, hourly and project-based, and you sign on as an independent contractor rather than an employee. Everything in this Mercor review applies to that expert side of the marketplace, not to the labs buying the work.

The company itself is not a side project. Investors valued Mercor at roughly $10 billion in 2026, and its founders rank among the youngest self-made billionaires on record, according to TechCrunch. A platform with that much money behind it far more reliably pays the people doing the work.

Mercor works best for people who already have something specific to offer. Think a practising physician, a qualified lawyer, a financial analyst, a senior developer or a native speaker of a language the labs are short on.

It works far less well as a first remote job. Entry-level tasks exist, but they are the lowest paid tier and the competition for them is enormous. If you are searching for AI training jobs with no experience at all, platforms that hire in volume will take you faster.

In addition, this suits people who can tolerate uneven income. Some contractors bill 40 hours a week for months. Others pass the interview and hear nothing.

The path from signing up to your first payout runs through four steps, and each one carries a wait. This Mercor review follows that path in order, so you know what happens at every stage.

You create an account, upload a CV or write a professional summary, and the platform builds a profile from it. After that comes the part everyone asks about: the Mercor AI interview.

It runs roughly 20 minutes on video, with no human on the other side. The system asks about your background and your domain, then follows up on your answers. Typical Mercor interview questions are not brain teasers. They ask what you have actually done, how you would judge a good answer in your field, and how you would turn a model into an expert in your domain.

You generally get up to three attempts. Because of that, treat the first one as practice. Speak in specifics, name real projects, and avoid the vague language that a screening model cannot score.

Are Mercor jobs real? Yes, and they are less glamorous than the marketing suggests. Most of the work is reinforcement learning from human feedback, which in practice means reading two model answers and explaining which is better and why.

Beyond that, you may write prompts, grade code, fact-check medical or legal output, or record reasoning in your own words so a model can learn from it. Some listings ask for 10 to 20 hours a week. Others want 40 or more.

Everything is remote and mostly asynchronous. You work against a weekly hour target, at whatever time of day suits you.

Each contract sets its own rate, and those rates swing wildly by domain. For this Mercor review we compared the published ranges side by side.

Independent trackers put AI lab contracts at roughly $30 to $150 an hour, with senior coding evaluators around $85 to $110. So the headline $200 figure is real but rare, and it belongs to credentialed specialists rather than generalists.

Contracts run in US dollars. Mercor releases payouts weekly, on Wednesday, covering the previous week’s hours, though some contracts follow a different cadence.

The money moves through Stripe in most countries and through Wise elsewhere. Stripe holds your very first payout for about seven days, then the weekly rhythm starts. From there, funds reach your bank in two to five days, or in around 30 minutes if you pay the 1% instant payout fee.

Monitoring software tracks your hours rather than the honour system, so log time as you work instead of reconstructing it later.

This is where most people give up. Hearing back on your application status commonly takes two to four weeks, and the stretch from applying to a first paid task is often three to six weeks.

Counting the held first payout, five to seven weeks between signing up and seeing money is normal. Consequently, nobody should treat Mercor as this month’s rent. Apply, then carry on with whatever you were already doing.

The core promise holds up. Contractors get paid, on time, at rates most freelance platforms cannot match, and the weekly rhythm is reliable enough that people plan around it.

What any honest Mercor review has to add is that the work arrives in waves. A lab collects the data it needs, then closes the project, sometimes earlier than the timeline contractors expected. In one widely reported case a lab cancelled a large project and later invited the same workers onto a similar one at a lower rate than the $21 an hour they had been earning. Others describe quiet weeks in the middle of a contract.

The screening also stays genuinely selective, and it rarely explains a rejection. Treat a no as a routing decision about which experts a lab needs this quarter, because that is mostly what it is.

On Trustpilot the picture runs mixed but broadly positive, around four stars across more than 200 reviews. The praise repeats one point: payments arrive accurately and on time. The complaints repeat three others, namely unpaid assessments, silence after passing, and work that dries up.

Plan around it the way you would plan around seasonal freelance work, and it holds up well.

Is Mercor free? Yes, for the expert side. You are the supply, not the customer, so creating a profile, sitting the interview and taking contracts costs nothing. The AI labs pay Mercor. Here is what the money side looks like for you.

What it costs you is time. Between the interview, the assessments and any unpaid trial task, expect to spend a few hours before you know whether the platform will use you at all. For this Mercor review we checked every figure against the platform’s own documentation and current contractor reports.

Most people who fail here fail at the profile stage, not at the expertise stage. A few habits separate the applicants who land a match quickly from the ones who wait.

Put your credentials where the matching can actually see them. Board certification, bar admission, the languages you work in, the frameworks you ship in production. A profile that says senior engineer competes with thousands. One that says distributed systems in Go, eight years in payments, competes with a handful.

Use the retakes. You generally get up to three attempts at the interview, so treat the first as a rehearsal, notice where you rambled, and come back tighter.

Answer with examples instead of adjectives. Name the case, the project, the decision you made and the reason behind it. A screening model cannot score enthusiasm, though it reads concrete detail very well.

Finally, apply to several listings and reply quickly when a match arrives. Roles open and close as labs start projects, and the seat usually goes to whoever is ready that day.

Is Mercor legit, or a scam? It is legit where it counts: the company is real, the contracts are real, and the money reaches people. It has a named leadership team, verified venture funding, public financial reporting and a large body of contractor feedback confirming that payouts land.

The scam accusations usually trace back to three friction points rather than to fraud. First, unpaid assessments feel like free labour when nothing follows. Second, silence after a passed interview reads as a con to anyone used to normal hiring. Third, cancelled projects hit contractors who had planned their month around the hours.

Security belongs in an honest answer too. In March 2026 attackers hit LiteLLM, an open-source component that runs right across the AI industry, and the compromised versions exposed candidate data at several companies including Mercor. The company emailed affected people within the week, reported the incident to the California Attorney General, and the resulting claims continue in court, as TechCrunch reported. Identity verification is a fixed cost of this industry, since the labs paying for the work require it, so take the usual step after any breach and freeze your credit file if you are in the US.

None of that is theft. Everything we found while researching this Mercor review points at the same risk, and that risk is silence rather than fraud.

Outlier, run by Scale AI, draws the obvious comparison in every Mercor review. It pays less, with RLHF work around $30 to $42 an hour and senior coding evaluation closer to $65 to $75. Onboarding, however, moves much faster, often days rather than weeks.

Most contractors who have used both run them side by side. Outlier covers the steady hours while Mercor stays open for the higher-paid specialist contracts.

DataAnnotation pays roughly $15 to $40 an hour on a task-by-task basis with weekly payments, and it accepts generalists. Compared with Mercor the ceiling is far lower, but the barrier is lower too.

If your goal is simply to get paid to train AI from home without a specialist credential, this is usually the faster route in.

Scale AI works with labs directly at enterprise level and reaches individuals mainly through Outlier. Micro1 is closer to Mercor’s model, screening engineers with AI interviews for contract placements, but it is smaller and more focused on software roles.

Registering with two or three of them costs you nothing extra and smooths out the gaps when one project ends.

Pros

• Genuinely high rates for credentialed experts, with medicine, law and finance reporting $75 to $200 an hour
• Weekly payouts in US dollars through Stripe or Wise, with a strong record of paying on time
• Fully remote, asynchronous contracts you can fit around an existing job
• Free to join, with no platform fee taken out of your hourly rate
• One 20-minute AI interview replaces the usual multi-round hiring process

Cons

• A March 2026 supply chain attack on a third-party open-source component exposed candidate data, and the resulting claims are still in court
• Selective screening with little feedback, so many applicants never learn why they were rejected
• Projects can be paused or cancelled with almost no notice, and rates on replacement work may be lower
• Five to seven weeks is normal between signing up and your first payment
• Assessments and short work trials are unpaid
• Entry-level rates of $12 to $25 an hour are ordinary compared with the headline figures

If you hold a real credential in medicine, law, finance or senior engineering, Mercor is one of the best-paid remote options open right now. Very little else routinely pays three figures an hour for work you can do from a laptop at 11pm, and the money arrives every week without chasing.

Applying costs you an evening. That is the whole downside for most people, because the interview runs short and the profile you build stays reusable. Even if the first match takes a month, your expertise sets the rate, not a freelance marketplace race to the bottom.

Two things keep the score in this Mercor review at three and a half rather than higher: the work arrives in project-shaped waves, and the March security incident is still on the record.

Our take: apply if your expertise is the product, keep a second platform running for steady hours, and let the Mercor contracts be the month that pays for something you actually wanted.

If the AI job market is what brought you here, our Jack & Jill AI review covers the platform that works the other side of the same problem, matching candidates and recruiters automatically.

And if you are weighing up which skills the labs will still be paying for next year, read Gen Z AI job training before you pick a direction.