Skip to content

Research report ยท Live data

What Raises AI Training Pay

Compared within each kind of work on 17 Oct 2026, a US-only location rule went with +55.2%, a professional credential in the description went with +21.7%, a profession named in the title went with +11.0% and a language named in the title with -20.8% in advertised pay.

Data window
20 Sep 2026 to 17 Oct 2026
28 days, UTC
Computed
18 Oct 2026, 17:20 UTC
Recomputed every day
Source
The public job boards of the AI training platforms this site tracks, read every day. Descriptions come from the platforms whose boards publish them in full.
Method
Version 1. Definitions
CC BY 4.0, how to cite

Every figure on this page as CSV and JSON, sent to your work email.

Key findings

  1. 1

    Within the same kind of work, roles open only to US residents advertised 55.2% more than roles open worldwide, the largest premium measured; roles whose title names a language advertised 20.8% less than roles whose title names none.

    Figure 1
  2. 2

    Roles whose title names a profession or discipline advertised 11.0% more than roles whose title names none in the same kind of work, across 10 kinds (medians $65 and $50/hr overall).

    Figure 2
  3. 3

    Roles whose title names a language advertised 20.8% less than roles whose title names none in the same kind of work, across 9 kinds (medians $40 and $65/hr overall).

    Figure 3
  4. 4

    Roles open only to US residents advertised 55.2% more than roles open worldwide in the same kind of work, across 11 kinds (medians $90 and $65/hr overall).

    Figure 4
  5. 5

    Among roles with a full description, a professional credential went with +21.7% in advertised pay within the same kind of work, and a PhD with -9.7%: a licence or professional qualification is priced above a doctorate.

    Figure 1

Figure 1

Pay premiums, compared within each kind of work

Which traits of a role come with higher advertised pay, all else in its kind of work equal?

TraitPremium within kindMedian withMedian without
Open only to US residents
+55.2%
$90$65
Professional credential in the description
+21.7%
$83$60
Title names an expert field
+11.0%
$65$50
Senior, staff, principal or lead title
-5.6%
$65$65
Assessment or test in the description
-8.1%
$65$65
PhD in the description
-9.7%
$65$65
Title names a language
-20.8%
$40$65

Sample. Roles open on 17 Oct 2026 that state an hourly rate. The last three traits compare the 3,260 open roles whose platform publishes a full description (AfterQuery, Anyone AI, Appen, Babel Audio, Braintrust, Claru, Handshake AI, Invisible, Mercor, micro1, Mindrift, OneForma, OpenTrain AI, Outlier (Scale AI), Pasiflora AI, Perle Labs, Pila8, RemoExperts, Remotasks, Sigma AI, Sovrano AI, SuperAnnotate (SME), Terac, TransPerfect DataForce, Turing, Vetto, Welo Data, xAI (SpaceXAI)).

Premium within kind: for each kind of work with at least 5 priced roles on both sides, the median rate with the trait over the median without it, minus one; then the average across those kinds, weighted by the roles with the trait. The medians with and without are across all kinds, for context: where they disagree with the premium, the trait is concentrated in better or worse paid kinds of work.

A premium is an association in advertised pay, not a cause: traits travel together (a credentialed role is often also senior).

Source. AITraining.jobs Research, What Raises AI Training Pay, live data.

Figure 2

Naming an expert field, kind by kind

Does a title that names a profession pay more within each kind of work?

Kind of workPremiumMedian withMedian without
Law, Medicine & Finance
+1.4%
$71$70
Coding & Software Evaluation
+8.3%
$65$60
Transcription, Audio & Speech
+66.7%
$50$30
Math & STEM
0.0%
$65$65
AI Evaluation & Rating (RLHF)
+44.4%
$65$45
Agentic Tasks & RL Environments
+26.3%
$60$48
Multilingual & Localization
-18.5%
$53$65
Data Annotation & Labeling
+42.9%
$40$28
Red-Teaming & AI SafetySmall sample
+7.9%
$72$67
Robotics & Physical-AI DataSmall sample
+25.0%
$13$10

Sample. Roles whose title names a profession or discipline, against roles whose title names none, open on 17 Oct 2026; kinds of work with at least 5 priced roles on both sides.

Rows with fewer than 10 priced roles on either side are marked: their medians move with a few postings.

Source. AITraining.jobs Research, What Raises AI Training Pay, live data.

Figure 3

Naming a language, kind by kind

Do roles for a named language pay more or less than other roles of the same kind?

Kind of workPremiumMedian withMedian without
Multilingual & Localization
-15.4%
$55$65
Transcription, Audio & Speech
-40.0%
$30$50
Math & STEM
0.0%
$65$65
Law, Medicine & Finance
-9.7%
$65$72
Data Annotation & Labeling
-49.2%
$20$40
Coding & Software Evaluation
0.0%
$65$65
AI Evaluation & Rating (RLHF)
-66.2%
$20$60
Research Studies & SurveysSmall sample
+81.8%
$100$55
Writing & ContentSmall sample
-45.7%
$31$58

Sample. Roles whose title names a language, against roles whose title names none, open on 17 Oct 2026; kinds of work with at least 5 priced roles on both sides.

Rows with fewer than 10 priced roles on either side are marked: their medians move with a few postings.

Source. AITraining.jobs Research, What Raises AI Training Pay, live data.

Figure 4

The US-only premium, kind by kind

Do roles open only to US residents pay more than roles open worldwide, within each kind of work?

Kind of workPremiumMedian withMedian without
Law, Medicine & Finance
+14.3%
$80$70
Multilingual & Localization
+150.0%
$100$40
Coding & Software Evaluation
-5.9%
$68$72
Data Annotation & Labeling
-20.0%
$34$43
Transcription, Audio & Speech
-21.1%
$28$36
AI Evaluation & Rating (RLHF)
-32.1%
$45$66
Math & STEM
-9.1%
$75$83
Robotics & Physical-AI DataSmall sample
-14.3%
$15$18
Agentic Tasks & RL EnvironmentsSmall sample
-20.8%
$48$60
Writing & ContentSmall sample
-5.9%
$60$64
Red-Teaming & AI SafetySmall sample
+90.0%
$95$50

Sample. Roles open only to US residents, against roles open worldwide, open on 17 Oct 2026; kinds of work with at least 5 priced roles on both sides.

Rows with fewer than 10 priced roles on either side are marked: their medians move with a few postings.

Source. AITraining.jobs Research, What Raises AI Training Pay, live data.

For labs, data vendors and researchers

Buy the data behind this report

The figures above are a summary. Behind them are records we collect ourselves, every day:

  • 9,572 roles tracked, each with its history

    Every role on 29 job boards, read daily since 13 Jul 2026: its pay and place, and the first and last day it was open.

  • 6,200 apply clicks in 28 days

    Which roles people actually apply to, by field, pay and kind of work. It happens on this site, so nobody else can count it.

  • 39 lab and vendor hiring boards

    The job boards of 25 AI labs and 14 human-data vendors, read every day: who is hiring for human data, and who is hiring to buy it.

What you can buy

  • The records

    The rows behind any figure, with their history, not only the totals.

  • A regular feed

    The same figures every week or month, as files.

  • A custom cut or benchmark

    Your countries, languages or fields, or your own platform against the market.

Method and definitions

The public job boards of the AI training platforms this site tracks, read every day. Descriptions come from the platforms whose boards publish them in full. Every figure is computed by one database query over the window shown, the same query for the live report and for every edition. Each figure states its sample beneath it.

Premium within kind
For each kind of work with at least 5 roles stating an hourly rate on both sides, the median rate with the trait over the median without it, minus one; averaged across those kinds, weighted by the roles with the trait. A trait needs two such kinds of work to show a premium.
Expert field, language
Named in the title, matched as whole words against the same vocabularies as the expert-fields and languages reports.
Professional credential
A description naming a medical, law, finance or nursing qualification, a board certification or a licence, read with the patterns published in the requirements report.
US-only
A role whose location rule names the United States and no other country, compared with roles open worldwide.
Advertised pay
The hourly figure a posting states (the middle of a stated range). Medians need at least 8 roles stating a rate; the middle half runs from the 25th to the 75th percentile.
Kind of work
Each role is classified from its title and description into one of the twelve kinds of AI training work this site tracks, from AI evaluation and coding to multilingual and audio work. Every kind links to its own page.
Open role
A role listed by a given day and still in its platform's feed that day.

What this data cannot tell you

  • Platforms that recruit without a public job board are not counted. Several large ones invite workers to projects directly.
  • A listing is not a hire. Some platforms keep evergreen roles open for months, and some post one listing per language or country.
  • A premium is an association in advertised pay, not a cause: traits travel together, and the comparison holds only the kind of work fixed.
  • Descriptions are published by a few platforms only, so the credential, PhD and assessment premiums describe those platforms.
  • Pay is what postings advertise, not what workers are paid after bonuses, tiers or rejected tasks.

How the job feeds are read and roles are classified: site methodology. How fresh, complete and consistent the data is, checked every day: data quality and coverage.

Editions

The live report covers the last 28 complete days and changes every day. Each complete month is also published as an edition, frozen two days after the month ends, so a figure you cite from an edition never moves.

  • Live dataLast 28 complete days, recomputed daily

Cite and reuse

Free to quote, chart and republish under CC BY 4.0 with a link to this page. Live figures change daily: cite a monthly edition for numbers that will read the same later, or keep the access date.

Citation

AITraining.jobs (2026). What Raises AI Training Pay (live data, 20 Sep 2026 to 17 Oct 2026). https://www.aitraining.team/research/ai-training-pay-premiums (accessed 18 Oct 2026)

BibTeX

@misc{aitrainingjobs_ai_training_pay_premiums_live,
  author = {{AITraining.jobs}},
  title = {What Raises AI Training Pay (live data, 20 Sep 2026 to 17 Oct 2026)},
  year = {2026},
  howpublished = {\url{https://www.aitraining.team/research/ai-training-pay-premiums}},
  note = {Accessed 18 Oct 2026},
}
Data: every figure as CSV (one line per value) and JSON, sent to your work email.Pay and open-roles data

More research

All 78 research reports