AI training jobs: every major platform, with sourced pay
Paid expert work that improves AI models: writing prompts and model answers, grading outputs, building rubrics, red-teaming. Hourly, remote, gated by tests.
13 checked options in this category, plus 2 closed or not checked yet. Realistic pay runs from $5–225/h across the 13 options with an hourly figure. Fastest first money: DataAnnotation, from 3 days. 13 of 13 cost nothing to start.
Best AI training platforms right now
Ranked by a published formula, not by opinion or payments. See How we rank.
- 82/100 Ethos pay 100 · grade B · speed 39 · status 100
- 71/100 Mercor pay 54 · grade A · speed 78 · status 60
- 68/100 AfterQuery pay 67 · grade B · speed 39 · status 100
- 65/100 micro1 pay 58 · grade B · speed 39 · status 100
- 56/100 Handshake AI Fellowship pay 50 · grade B · speed 39 · status 60
The formula
Score out of 100 = 40% realistic hourly pay (the middle of the range, scaled from the lowest to the highest in this list) + 25% evidence grade (A 1, B 0.75, C 0.5, D 0.25) + 20% speed to first money (fewest days scores highest, scaled within the list) + 15% status (open 1, turbulent 0.6, waitlist 0.5, pilot 0.5, unverified 0.3, paused 0.1). Unchecked and closed platforms are left out.
Compared side by side
How this list is sorted: best documented first (evidence grade, then the number of sourced facts). Payments never change the order. How we rank How well-sourced a page or fact is. Not a rating of the company. More in the glossary
| Name | Step | Realistic pay | First money | Cost to start | Status | Grade | Checked |
|---|---|---|---|---|---|---|---|
| Mercor | 1 · Paid tasks | ≈ $16–150/h | 7–60 days | $0 | Turbulent status, as of 13 Oct 2026 | Grade A: Official source Grade A means: Official source. The grade says how well-sourced this page is, not how good Mercor is. More in the glossary | |
| Outlier | 1 · Paid tasks | ≈ $12–60/h | 7–30 days | $0 | Turbulent status, as of 13 Oct 2026 | Grade 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 | |
| DataAnnotation | 1 · Paid tasks | ≈ $14–45/h | 3–30 days | $0 | Turbulent status, as of 13 Oct 2026 | Grade B: Reliable source Grade B means: Reliable source. The grade says how well-sourced this page is, not how good DataAnnotation is. More in the glossary | |
| Ethos | 1 · Paid tasks | ≈ $60–225/h | 14–60 days | $0 | Open status, as of 13 Oct 2026 | Grade 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 | |
| Handshake AI Fellowship | 1 · Paid tasks | ≈ $30–125/h | 14–45 days | $0 | Turbulent status, as of 13 Oct 2026 | Grade 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 | |
| micro1 | 1 · Paid tasks | ≈ $25–150/h | 14–60 days | $0 | Open status, as of 13 Oct 2026 | Grade 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 | |
| AfterQuery | 1 · Paid tasks | ≈ $50–150/h | 14–45 days | $0 | Open status, as of 13 Oct 2026 | Grade 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 | |
| TELUS Digital AI Community | 1 · Paid tasks | ≈ $8–17/h | 21–60 days | $0 | Open status, as of 13 Oct 2026 | Grade B: Reliable source Grade B means: Reliable source. The grade says how well-sourced this page is, not how good TELUS Digital AI Community is. More in the glossary | |
| Alignerr | 1 · Paid tasks | ≈ $15–60/h | 14–45 days | $0 | Turbulent status, as of 13 Oct 2026 | Grade C: Self-reported Grade C means: Self-reported. The grade says how well-sourced this page is, not how good Alignerr is. More in the glossary | |
| Mindrift | 1 · Paid tasks | ≈ $12–60/h | 14–45 days | $0 | Open status, as of 13 Oct 2026 | Grade C: Self-reported Grade C means: Self-reported. The grade says how well-sourced this page is, not how good Mindrift is. More in the glossary | |
| Turing | 1 · Paid tasks | ≈ $25–90/h | 14–28 days | $0 | Open status, as of 13 Oct 2026 | Grade C: Self-reported Grade C means: Self-reported. The grade says how well-sourced this page is, not how good Turing is. More in the glossary | |
| Appen / CrowdGen | 1 · Paid tasks | ≈ $5–20/h | 21–90 days | $0 | Open status, as of 13 Oct 2026 | Grade C: Self-reported Grade C means: Self-reported. The grade says how well-sourced this page is, not how good Appen / CrowdGen is. More in the glossary | |
| Meridial | 1 · Paid tasks | ≈ $8–50/h | 14–45 days | $0 | Open status, as of 13 Oct 2026 | Grade C: Self-reported Grade C means: Self-reported. The grade says how well-sourced this page is, not how good Meridial is. More in the glossary |
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.
Not in the table: SME Careers (not checked yet), Stellar AI (closed).
Compare and find alternatives
Side by side Mercor vs OutlierDataAnnotation vs OutlierMercor vs micro1Alignerr vs OutlierHandshake AI Fellowship vs OutlierDataAnnotation vs Mercormicro1 vs OutlierMercor vs TuringAlignerr vs MercorMindrift vs OutlierAppen / CrowdGen vs TELUS Digital AI CommunityHandshake AI Fellowship vs MercorDataAnnotation vs micro1DataAnnotation vs Handshake AI FellowshipHandshake AI Fellowship vs micro1Ethos vs Mercor
Alternatives Alternatives to MercorAlternatives to OutlierAlternatives to DataAnnotationAlternatives to micro1Alternatives to AlignerrAlternatives to Handshake AI Fellowship
More on AI training pay and risks
AI Trainer Pay Index: advertised vs. real pay Lawsuits and legal cases at AI training platforms
Related guides
- Earn now, build next · Why ways to earn online sit on one ladder, from paid tasks to owning a business, and how to pair income you can get this month with something you build for later.
- How to get accepted on AI training platforms · What Mercor, micro1, Outlier, DataAnnotation, Handshake AI and Turing ask for, which CV rules are official and which are only rules of thumb, and how to prepare for AI interviews.
Other categories
Microtasks, studies & user testingFreelance marketplaces & talent networksExpert networks & paid consultationsTutoring, teaching & coursesLanguage servicesVoice, likeness & content licensingCreator & audience monetizationDigital products & storefrontsSoftware, apps & micro-SaaSAI services & productized agenciesData, directories & media assetsAgent-economy playsHigh-leverage AI careersCommerce, physical products & offlineBuying & owning businesses