2026 pay reference
AI training salary guide 2026
Generalist AI training work pays $15–$40/hr. Coding and STEM tracks run $30–$75/hr. Credentialed experts in medicine, law, finance, and advanced STEM reach $50–$200/hr. The track you qualify for sets the band, not the title on the listing.
Pay by role
Bands are gross contractor rates in USD. They come from platform-published 2026 figures and live NearSkill listings. A specialist band is not a guarantee; it is what the track pays when the work is available.
- Entry
- $20–$40/hr
- Specialist
- $30–$75/hr
- Expert
- $50–$150/hr
Generalist ranking and writing; coding and STEM sit in the middle.
- Entry
- $15–$25/hr
- Specialist
- $30–$75/hr
- Expert
- Rare; usually moves into evaluation
Lowest barrier. Coding and specialist annotation pay the most.
- Entry
- $22–$40/hr
- Specialist
- $40–$70/hr
- Expert
- $50–$150/hr
Written justification is the product; expert tracks need a credential.
- Entry
- $15–$40/hr
- Specialist
- $40–$75/hr
- Expert
- $50–$150/hr
Language and STEM explanation work; credentialed subjects pay more.
- Entry
- $25–$50/hr
- Specialist
- $60–$150/hr
- Expert
- $100–$200/hr
Generalist coding is crowded; security and ML depth raise the ceiling.
- Entry
- $15–$25/hr
- Specialist
- $20–$40/hr
- Expert
- $40+/hr
Scarce language pairs pay more than common ones.
- Entry
- Credential required
- Specialist
- $50–$150/hr
- Expert
- $100–$200/hr
Medicine, law, finance, and advanced STEM. Verification is the gate.
| Role | Entry | Specialist | Expert | What moves it |
|---|---|---|---|---|
| AI trainer | $20–$40/hr | $30–$75/hr | $50–$150/hr | Generalist ranking and writing; coding and STEM sit in the middle. |
| Data annotator | $15–$25/hr | $30–$75/hr | Rare; usually moves into evaluation | Lowest barrier. Coding and specialist annotation pay the most. |
| AI evaluator | $22–$40/hr | $40–$70/hr | $50–$150/hr | Written justification is the product; expert tracks need a credential. |
| AI tutor | $15–$40/hr | $40–$75/hr | $50–$150/hr | Language and STEM explanation work; credentialed subjects pay more. |
| AI code reviewer | $25–$50/hr | $60–$150/hr | $100–$200/hr | Generalist coding is crowded; security and ML depth raise the ceiling. |
| AI linguist | $15–$25/hr | $20–$40/hr | $40+/hr | Scarce language pairs pay more than common ones. |
| AI domain expert | Credential required | $50–$150/hr | $100–$200/hr | Medicine, law, finance, and advanced STEM. Verification is the gate. |
Pay by platform
Platform bands overlap because they price different tracks. A wide band does not mean every worker qualifies for the top. Check the schedule column before the rate column, because slow payment changes the real value of a high rate.
$50–$150/hr expert; $16–$100+/hr generalist
Payment: Weekly
Highest expert ceiling; frontier-lab projects
$15–$150/hr; median about $27/hr
Payment: Biweekly or monthly
Steady contract volume; US Pacific overlap
$15–$130+/hr by expertise
Payment: Weekly via Deel
Credentialed expert projects in regulated fields
$60–$220/hr by expert network
Payment: On verified completion
Highest published engineering and ML bands
$25–$250/hr; per-task work separate
Payment: Weekly via Stripe
Credentialed reasoning and evaluation work
$17–$125/hr; expert floor $60/hr
Payment: Weekly via Stripe
Highest verified expert floor; US-only
$6–$150/hr by track and location
Payment: Every two weeks
Wide marketplace; location affects the offer
| Platform | Published band | Payment | Strength |
|---|---|---|---|
| Mercor | $50–$150/hr expert; $16–$100+/hr generalist | Weekly | Highest expert ceiling; frontier-lab projects |
| Micro1 | $20–$150/hr by track | Biweekly | Strong payment record; clear engineering track |
| Turing | $15–$150/hr; median about $27/hr | Biweekly or monthly | Steady contract volume; US Pacific overlap |
| SME Careers | $15–$130+/hr by expertise | Weekly via Deel | Credentialed expert projects in regulated fields |
| Terac | $60–$220/hr by expert network | On verified completion | Highest published engineering and ML bands |
| AfterQuery | $25–$250/hr; per-task work separate | Weekly via Stripe | Credentialed reasoning and evaluation work |
| Handshake AI | $17–$125/hr; expert floor $60/hr | Weekly via Stripe | Highest verified expert floor; US-only |
| Meridial | $6–$150/hr by track and location | Every two weeks | Wide marketplace; location affects the offer |
The headline rate is not your rate
Every band above is gross pay for paid time. The number you actually keep is lower. Unpaid onboarding, assessment time, idle queues, rejected tasks, and self-employment tax all come out of the same hour. A $40/hr project with five unpaid hours a week and two empty days produces an effective rate closer to $22–$25/hr.
- 1. Log paid and unpaid hours separately for one month.
- 2. Divide gross pay by total hours, including onboarding and idle time.
- 3. Subtract platform fees and your tax reserve.
- 4. Compare that number across platforms, not the advertised band.
The one lever that moves pay
Track assessments, not years of experience, set the band. A generalist who passes a coding or expert assessment can double their rate on the same platform. A certificate does not move the band unless the platform recognizes it. The assessment is the gate.
Frequently asked questions
How much do AI training jobs pay in 2026?
Generalist AI training work pays about $15–$40/hr. Coding and STEM tracks run $30–$75/hr. Credentialed expert work reaches $50–$200/hr, with medicine, law, finance, and advanced STEM at the top. The rate tracks the assessment track and the credential, not years of general experience.
Do AI training jobs pay hourly or per task?
Both. Most platforms publish hourly bands, and some projects pay a fixed amount per task or batch. A $300 per-task rate is not comparable to $100/hr unless you know the task takes three hours. Convert every per-task offer to an effective hourly rate before you accept.
Why is my effective rate lower than the advertised rate?
The advertised rate excludes unpaid onboarding, assessment time, idle queues, rejected tasks, and self-employment tax. A $30/hr band with five unpaid hours a week and two empty days can produce an effective rate near $15–$18/hr. Track your actual paid hours for a month before comparing platforms.
Which AI training platform pays the most?
At the top, Terac publishes engineering bands up to $220/hr and AfterQuery reaches $250/hr on rare specialist tracks. Mercor advertises $50–$150/hr for experts, and Handshake AI holds a $60–$100/hr credentialed band. The highest number is not always the best fit: eligibility, payment timing, and work stability decide the real return.
Can beginners earn $100/hr in AI training?
Rarely. Generalist tracks pay $15–$40/hr, and expert bands require a verified credential or a senior record. The realistic path is a generalist or annotation start, then a track assessment, then an expert application after you have documented work. Anyone promising $100/hr to beginners is selling something.
How do I raise my AI training rate?
Pass a harder track assessment, add a verifiable credential, move from per-task to hourly expert work, and apply to platforms with higher published bands. Switching platforms without changing track usually produces the same rate. The track is the lever.
Sources are linked from each role guide and platform page. Rates are USD contractor pay, indicative rather than quoted, and change with client demand. NearSkill does not estimate missing pay or merge hourly and per-task figures.
See the roles behind the bands
Browse live AI training, evaluation, and expert roles with the published rate on each listing.