Is Mercor legit? Pay, payouts and red flags, with sources
The largest expert marketplace for AI labs: hourly contracts for professionals, from generalists to doctors, lawyers and senior engineers.
Grade A: Official source. The grade says how well-sourced this page is, not how good Mercor is. How well-sourced a page or fact is. Not a rating of the company. More in the glossary
Updated Checked Checked: we looked again and it was still right. Updated: something changed. More in the glossary
Active, but turbulent Growing fast. In 2026 a reported data breach was followed by proposed class actions (pending), and large project offboardings were reported. As of 13 Oct 2026.
Mercor matches vetted people to project-based contracts for AI labs and enterprises (writing and grading model outputs, building rubrics and benchmarks, domain evaluations). Matching runs on your profile, an AI interview and assessments; a strong, complete profile can trigger 'instant offers' without applying. Work is contract-based and can end abruptly when a client project ends.
How much does Mercor pay?
Pay figures are what platforms advertise or what workers report, each with its date. They are not typical or guaranteed earnings: yours depend on your skills, hours, location and how much work is available.
Note Advertised 'average contracted rate' oscillated ~$80–141/h in 2026 (≈$108 on 2026-10-04). Realized pay depends on hour caps, project length and per-task rates.
- Advertised$20–250/h
- Median listed$75/haitrainer.work (P. Romeo), 12 Oct 20266
- Realistic range$16–150/h
- Advertised average contracted rate: $108/h (2026-10-04); range ~$80–141/h during 2026 5
- Listing pay band: Expert listings roughly $60–250/h (e.g., Spanish tutors $60–65, Software Engineer Expert $100–150, Cybersecurity Research $200–250) 5
- Median listed rate across live roles: $75/h (n=1,377 listings with hourly rate) 6
- Pay cut on project change: Meta-related review project ended Nov 2025; workers offered rehire at $16/h (from $21/h) 9
How and when does Mercor pay?
Who can join Mercor?
Where it's open: Worldwide, except Cuba, Iran, North Korea, Syria.
Regions: global (Stripe Connect countries; Wise as fallback) · Not in: sanctioned countries · Work authorization: verified per application · Some listings are region-locked; listings without a location default to US/CA/UK/EU (referral policy).
- Education
- varies by listing; top rates need graduate degree, license or senior industry experience
- Experience
- from generalist roles to 10+ years specialists
- Credentials
- MD/JD/CFA etc. for specialist tracks
- Equipment
- desktop or laptop with Chrome; camera + microphone; Insightful desktop app (no Chromebook/iPad/phone)
- Identity check
- Yes, with ID documents
- Minimum age
- 18
- Unpaid screening
- Tests or onboarding before paid work
- Monitoring
- Software that records your work
- Also
- Work authorization verified; legal name must match ID
Most platforms here require you to be 18 or older.
How to get accepted
- Write a keyword-rich summary and quantified bullets (see CV playbook)
- Upload a text-based PDF CV (no graphics, <2 MB)
- Set accurate availability, location and work authorization
- Do the AI interview on desktop Chrome in a quiet room; only the latest attempt counts
- Enable notifications so instant offers don't expire
- Resume format: PDF only, OCR-friendly text, 1–2 pages, <2 MB 4
Tips
- Retaking an AI interview resets that application's progress
- Hide your profile or request data deletion if you stop using it
- Treat each contract as temporary; keep 2–3 platforms warm
Guide: How to get accepted on AI training platforms Your CV decides a lot here: see how your skills score
Is Mercor legit?
What we can confirm
- Documented weekly payout process (Stripe, Wise)
- $10B valuation (Oct 2025); talks at $20B (Jul 2026)
- Public talent documentation incl. machine-readable llms-full.txt
- Referral terms: 20% of referee earnings up to 4x referee hourly rate (hourly contracts) 2
- Valuation & revenue: $10B (Oct 2025, $350M Series C); talks at $20B; run-rate >$2B (CEO, July 2026) 7
What to watch
- 2026: a reported data breach via a supply-chain attack on the LiteLLM library was followed by six proposed class actions (pending, allegations not proven). The platform collects ID documents and biometric data for verification, so the stakes of a breach are high
- Impersonation scams: real offers only arrive inside the Mercor platform
- Projects can end without notice; large offboardings were reported in 2026
- Pay can be cut on renewal (Nov 2025: $21/h -> $16/h on a rehire offer)
- Monitoring software takes random screenshots
- Long waits in the talent pool are common
- Data breach & lawsuits: Reported breach via a supply-chain attack on the LiteLLM library; 6 proposed class actions filed (pending); Meta reportedly paused work with Mercor 8
- Community signal (Reddit, July 2026): Breaking Even's tally: 519 posts; top tags 'actively hiring', 'work drought', 'account bans'; a large offboarding thread 10 Community-reported: self-reports and post counts from public forums, as dated. Not verified by us; individual experiences vary.
Lawsuits and complaints are allegations unless a court has ruled. We report their status with dates and sources and update it when it changes.
What people like
- Highest expert rates in the category
- Weekly pay via Stripe/Wise
- Passed assessments are reused across similar roles
- Instant offers can arrive without applying
Alternatives to Mercor
- Outlier ≈ $12–60/h · first money 7–30 days
- DataAnnotation ≈ $14–45/h · first money 3–30 days
- Ethos ≈ $60–225/h · first money 14–60 days
- Handshake AI Fellowship ≈ $30–125/h · first money 14–45 days
All 8 alternatives to Mercor, compared Mercor vs OutlierMercor vs micro1Mercor vs DataAnnotationMercor vs TuringMercor vs AlignerrMercor vs Handshake AI FellowshipMercor vs Ethos
The proof
Every fact on this page with its type, grade, date and source. Report an error
Sourced facts 11 facts · type · grade · as of
- From the company Payout schedule
Hours finalized Fri 11:59 PM IST; processed by end of Wednesday (PST); funds usually land Thursday 1
- From the company Payout methods
Stripe primary; Wise where Stripe is unavailable; first Stripe payout has a 7-day hold 1
- From the company Referral terms
20% of referee earnings up to 4x referee hourly rate (hourly contracts) 2
- Marketing claim Advertised average contracted rate
$108/h (2026-10-04); range ~$80–141/h during 2026 5
- Marketing claim Listing pay band
Expert listings roughly $60–250/h (e.g., Spanish tutors $60–65, Software Engineer Expert $100–150, Cybersecurity Research $200–250) 5
- Press & research Median listed rate across live roles
$75/h (n=1,377 listings with hourly rate) 6
Show 5 more facts
- Press & research Valuation & revenue
$10B (Oct 2025, $350M Series C); talks at $20B; run-rate >$2B (CEO, July 2026) 7
- Press & research Data breach & lawsuits
Reported breach via a supply-chain attack on the LiteLLM library; 6 proposed class actions filed (pending); Meta reportedly paused work with Mercor 8
- Press & research Pay cut on project change
Meta-related review project ended Nov 2025; workers offered rehire at $16/h (from $21/h) 9
- From the company Resume format
PDF only, OCR-friendly text, 1–2 pages, <2 MB 4
- Workers & founders say Community signal (Reddit, July 2026)
Breaking Even's tally: 519 posts; top tags 'actively hiring', 'work drought', 'account bans'; a large offboarding thread 10
Community-reported: self-reports and post counts from public forums, as dated. Not verified by us; individual experiences vary.
Sources 10 sources · type
- Mercor talent docs: Payouts Watched daily, last check , changed 14 Oct 2026
- Mercor talent docs: Referral policy Watched daily, last check , changed 14 Oct 2026
- Mercor talent docs: Why did I receive an instant work offer? Watched daily, last check , changed 14 Oct 2026
- Mercor talent docs: How to apply Watched daily, last check , changed 14 Oct 2026
Grade A (official source) · checked 13 Oct 2026 · re-checked every 14 days · how grades work · report an error
Changelog
Every change to this page, newest first. In the full changelog
- added all
Keep exploring
On the ladder
Step 1 of 5 · Paid tasks. The platform finds the work and sets the pay. Fastest first money, lowest ceiling.
Where this leads
- Founder content engine / AI-assisted ghostwriting Evaluating AI writing every day teaches exactly what separates slop from a person's real voice; companies pay for that.
- AI tools assessment for small businesses You know where models fail; small businesses need someone who can tell them which tools to trust.
- Expert networks Uses the same skills: analysis, medicine, law.
Related
- Step 1 · Paid tasks Open status
micro1
AI-interview-gated talent pool ('Zara') placing experts on AI-lab projects; strong rates for STEM, legal and healthcare.
Realistic pay ≈ $25–150/hFirst money 14–60 daysCost to start $0Grade B: Reliable source Grade B means: Reliable source. The grade says how well-sourced this page is, not how good micro1 is. More in the glossary Reliable source Checked 4 days ago - Step 1 · Paid tasks Turbulent status
Outlier (Scale AI)
Scale AI's contributor platform: rate, write and fix AI outputs, with specialist tracks up to ~$150/h advertised.
Realistic pay ≈ $12–60/hFirst money 7–30 daysCost to start $0Grade B: Reliable source Grade B means: Reliable source. The grade says how well-sourced this page is, not how good Outlier is. More in the glossary Reliable source Checked 4 days ago - Step 1 · Paid tasks Turbulent status
Handshake AI Fellowship
AI training projects for US-based students, grads and PhDs via the Handshake careers network.
Realistic pay ≈ $30–125/hFirst money 14–45 daysCost to start $0Grade B: Reliable source Grade B means: Reliable source. The grade says how well-sourced this page is, not how good Handshake AI Fellowship is. More in the glossary Reliable source Checked 4 days ago - Step 1 · Paid tasks Open status
Ethos
London-based AI expert network (a16z-backed): one voice interview builds your profile; matched to expert calls, surveys and AI-data work.
Realistic pay ≈ $60–225/hFirst money 14–60 daysCost to start $0Grade B: Reliable source Grade B means: Reliable source. The grade says how well-sourced this page is, not how good Ethos is. More in the glossary Reliable source Checked 4 days ago - Step 1 · Paid tasks Open status
AfterQuery
Fast-growing YC startup paying working professionals to document how they reason through real cases.
Realistic pay ≈ $50–150/hFirst money 14–45 daysCost to start $0Grade B: Reliable source Grade B means: Reliable source. The grade says how well-sourced this page is, not how good AfterQuery is. More in the glossary Reliable source Checked 4 days ago