AI training pay report
Data as of . Rebuilt daily from Rubric's job list and platform grade sheets. Methodology
As of November 6, 2026, the typical top rate on open AI training jobs is $70/hr: half of the listings that state an hourly rate offer that or more. Platforms' own advertised ranges run about $15–$105/hr, with science, medical and legal work at the top.
- Open listings tracked
- 1,815
- Platforms with open listings
- 10
- Listings with an hourly rate
- 1,713 94%
- Typical top rate
- $70/hr
Other ways to read the same data: the median midpoint of listed ranges is $60/hr, and the median platform advertises about $15–$105/hr. The typical top rate is the figure used everywhere on Rubric.
What does each AI training platform advertise per hour?
Each bar runs from the platform's advertised floor to its ceiling, in USD per hour. marks the typical top rate on its open listings today, for platforms with 10 or more priced listings (the same rule as the home page). means live listings go above the advertised range. Sorted by ceiling.
AfterQuery$40–250
MercorTypical top rate on open listings: $90/hr$20–200†
BraintrustTypical top rate on open listings: $31/hr$20–200
AlignerrTypical top rate on open listings: $75/hr$15–150†
DataAnnotationTypical top rate on open listings: $125/hr$20–150
micro1Typical top rate on open listings: $90/hr$15–135†
SME CareersTypical top rate on open listings: $50/hr$20–130†
Handshake AI$30–125†
Remo$25–105
xAI$35–100
Turing$12–90
MindriftTypical top rate on open listings: $31/hr$15–90†
InvisibleTypical top rate on open listings: $65/hr$6–65†
Outlier$10–50
Prolific$8–30
TELUS Digital$3–25
Appen CrowdGen$2–25
| Platform | Floor | Ceiling | Notes |
|---|---|---|---|
| AfterQuery | $40 | $250 | General expert contributors earn about $40–50/hr, specialists $80–150/hr, and top specialists up to $250/hr; listing median is in the high $80s. |
| Mercor | $20 | $200 | Most listings sit around $50–100/hr; healthcare, legal and finance roles pay the most, language and generalist roles the least. |
| Braintrust | $20 | $200 | Braintrust lists $20–45/hr for data labeling, $25–60/hr for RLHF rating and up to $200/hr for expert code review; listings average around $50/hr. |
| Alignerr | $15 | $150 | Generalist and bilingual work is around $15–20/hr, coding QA $30–40/hr, and senior software, advanced math, medical and law roles $70–150/hr; some projects pay per approved task. |
| DataAnnotation | $20 | $150 | Official starting rates are $25–50/hr general, $40–150/hr coding and $40–125/hr STEM/professional; workers report general tasks often settle around $20–30/hr. |
| micro1 | $15 | $135 | Generalist and annotation roles start around $15–30/hr; coding, STEM and law/medicine roles commonly list $40–90/hr and occasionally more. |
| SME Careers | $20 | $130 | Listings show only an 'up to' rate: about $20/hr for generalists, $60/hr software, $90/hr healthcare and $110/hr legal; your actual rate depends on credentials and location. |
| Handshake AI | $30 | $125 | Handshake lists $30–125/hr based on education level and project; physician roles have listed $200/hr, and some projects pay per task. |
| Remo | $25 | $105 | Each listing sets its own rate; generalist roles list around $25–50/hr and specialist medicine, law and finance roles pay the most. |
| xAI | $35 | $100 | Posted US rates: languages $35–45/hr, writing and design $35–75/hr, software and legal up to $100/hr; pay for non-US hires is shared during recruiting. |
 Snorkel AI | $10/task | $100/task | Paid per accepted task, not per hour: $10–100 per task depending on complexity, with STEM reasoning tasks at $100 each (capped at 5 a day). Rejected work is unpaid. |
| Turing | $12 | $90 | Turing does not publish rates; entry LLM-trainer roles open to India list about $12–30/hr while senior coding and expert roles reach $40–90/hr, and some projects pay per task. |
| Mindrift | $15 | $90 | Paid per approved task; Mindrift quotes about $30/hr for general projects and up to $90/hr for specialist technical work, and rejected or expired tasks are unpaid. |
| Invisible | $6 | $65 | Listings show $6–65/hr and the rate depends on your country; many entry projects pay around $15–25/hr. |
| Outlier | $10 | $50 | English generalist work is usually $15–25/hr; coding and expert projects list $30–50/hr and rare specialist ML roles have listed more. Some non-English language projects pay under $10/hr. |
| Prolific | $8 | $30 | Studies must pay at least $8/hr (recommended $12/hr); AI tasker and domain-expert work pays more, with listings up to about $30/hr. |
| TELUS Digital | $3 | $25 | Rater pay is set by country: about $14–17/hr in the US and roughly $3–8/hr in India; AI Community expert roles (coding, ML, legal) list $20–25/hr. |
| Appen CrowdGen | $2 | $25 | Rates depend mostly on country and language, not skill; US and Western Europe projects run about $10–25/hr while many projects elsewhere pay a few dollars an hour or per unit. |
Which AI training fields pay the most?
For each field, the median floor and median ceiling among the platforms that hire for it (fields with at least three platforms, so two platforms can't set the going rate).
What do open AI training jobs pay right now?
From the 1,815 open listings on Rubric on November 6, 2026. "Typical top" is the median of each listing's top hourly rate; low and high are the lowest floor and highest ceiling listed. Groups with fewer than 5 paid listings are marked as small samples.
| Platform | Open jobs | With pay | Typical top | Lowest floor | Highest top |
|---|---|---|---|---|---|
| DataAnnotation | 104 | 104 | $125 | $40 | $150 |
| Handshake AIsmall sample | 3 | 3 | $125 | $32 | $158 |
| Mercor | 364 | 320 | $90 | $10 | $450 |
| micro1 | 347 | 339 | $90 | $6 | $400 |
| Alignerr | 450 | 450 | $75 | $10 | $280 |
| Invisible | 184 | 178 | $65 | $5 | $150 |
| SME Careers | 264 | 263 | $50 | $3 | $150 |
| xAIsmall sample | 5 | 4 | $35 | $35 | $40 |
| Braintrust | 58 | 29 | $31 | $8 | $150 |
| Mindrift | 36 | 23 | $31 | $9 | $100 |
| Field | Open jobs | With pay | Typical top | Lowest floor | Highest top |
|---|---|---|---|---|---|
| Legal | 100 | 99 | $120 | $6 | $400 |
| Medical | 158 | 138 | $100 | $8 | $450 |
| Finance | 148 | 145 | $100 | $6 | $350 |
| Coding | 242 | 237 | $90 | $6 | $350 |
| Science | 114 | 106 | $90 | $10 | $400 |
| Math | 47 | 47 | $90 | $6 | $200 |
| STEM | 158 | 148 | $70 | $6 | $300 |
| Image | 80 | 47 | $70 | $6 | $140 |
| Writing | 87 | 80 | $68 | $6 | $350 |
| Generalist | 204 | 197 | $50 | $3 | $350 |
| Languages | 204 | 204 | $50 | $5 | $65 |
| Video | 28 | 27 | $50 | $3 | $120 |
| Audio | 245 | 238 | $40 | $6 | $280 |
Browse the listings behind these numbers on all open jobs, or only the newest in new this week. To turn any of these rates into monthly pay at the hours you can work, use the pay calculator.
How is this report calculated?
- Open listings come from Rubric's daily job collection: public listings on each platform's own job pages, refreshed every day. A listing counts once, under the platform that posts it. See how jobs are collected.
- Hourly only. Listings paid per task, per project or with no stated rate are counted in "Open jobs" but excluded from pay figures. We don't convert per-task pay to hourly, because time per task varies too much.
- Typical top rate is the median of each listing's top hourly rate (a $40–60/hr listing counts as $60). It's the one headline figure Rubric uses on every page. The midpoint median (a $40–60/hr listing counts as $50) is shown as a secondary figure. Lowest floor is the lowest stated floor; highest top is the highest stated ceiling. Platform pages quote the lowest and highest top rates instead, and say so.
- Advertised ranges are the floor and ceiling each platform publishes or shows on its listings, recorded on our platform grade sheets with sources, and re-checked at least monthly.
- Field medians take the median of platform floors and ceilings across platforms that hire for that field. They describe what platforms advertise, not what an average worker earns.
- Currency: USD. Many platforms pay less in lower-cost countries; these numbers are not adjusted by country.
- What this is not: it isn't a survey of actual earnings. Real income depends on hours available, which is the biggest variable in this work. See realistic monthly earnings.
Cite as: Rubric AI Training Pay Report, data as of 2026-11-06. https://www.rubric.world/pay-report
Other questions people ask
How much do AI training jobs pay per hour?
Across the 17 platforms we grade that pay by the hour, the median advertised range is about $15–$105/hr. On open listings, the typical top rate is $70/hr (1,713 listings with an hourly rate); using the midpoint of each range instead, the median is $60/hr. Generalist work is usually at the low end; credentialed experts at the high end.
Which AI training field pays the most?
By advertised ceiling, science, medical, legal pay the most, with median ceilings of $130/hr, $130/hr, $130/hr among platforms hiring for them.
How often is this report updated?
The job numbers rebuild every day from the listings on Rubric. Platform pay ranges are re-checked when platforms change them and at least monthly. The date at the top shows when the data was pulled.
Can I cite this report?
Yes. Please cite it as "Rubric AI Training Pay Report, data as of 2026-11-06" and link to https://www.rubric.world/pay-report.