Report
AI training market report
A live view of AI training and evaluation work currently tracked by Sidequest: open roles, advertised hourly rates, hiring fields and platform activity.
Data scope
Sidequest tracks publicly listed AI training and evaluation opportunities across multiple platforms. The figures below describe the current Sidequest dataset, not the entire global AI labour market.
Figures are generated from the current Sidequest inventory when this page is rendered · computed 16 September 2026 · counts are published listings only, never profile entries or application pools · rates are advertised listing rates, not earnings. Methodology
Open listings
1255
Published vacancies currently tracked. Profile entries and pools excluded.
Platforms with listings
5
Of 6 platforms on the board; the rest are profile entries.
Median advertised rate
$78/hr
Median of 1007 hourly listings with a positive band.
Highest advertised rate
$400/hr
The top of the highest advertised band. A ceiling one listing states, not expected pay.
Listings with hourly rate data
80.2%
1007 of 1255 listings are hourly with a positive band.
Largest field
Languages
202 listings, 16.1% of the dataset.
Among listings currently tracked by Sidequest, 1255 are open vacancies across 5 platforms. 1007 of them advertise an hourly band with a positive floor, and the median of those bands is $78/hr. Every figure is recomputed from the live board on each visit; see the rates page for the same medians by field, or the board for the listings themselves.
AI training jobs by field
Every current listing, once, under the field Sidequest classified it in. Shares are of all 1255 listings; a median is shown once a field has 3 hourly listings with a positive band, the same floor the rates page applies.
| Field | Listings | Share of dataset | Median advertised rate | Highest advertised rate | Hourly listings |
|---|---|---|---|---|---|
| Languages | 202 | 16.1% | $30/hr | $300/hr | 181 |
| Generalist | 199 | 15.9% | $80/hr | $280/hr | 159 |
| STEM and Coding | 180 | 14.3% | $80/hr | $280/hr | 107 |
| Sciences | 149 | 11.9% | $80/hr | $280/hr | 107 |
| Other | 123 | 9.8% | $78/hr | $250/hr | 88 |
| Finance | 105 | 8.4% | $100/hr | $300/hr | 98 |
| Medical | 103 | 8.2% | $88/hr | $400/hr | 96 |
| Legal | 96 | 7.6% | $115/hr | $400/hr | 87 |
| Creative | 83 | 6.6% | $55/hr | $300/hr | 69 |
| Education | 15 | 1.2% | $60/hr | $120/hr | 15 |
AI training jobs by platform
Sorted by current listing count, which is the only ordering here — nothing is ranked or rated. A platform whose board presence is a profile entry shows no count, because a profile is not a listing. Inventory changes as platforms publish or close opportunities.
| Platform | Listings | Share of dataset | Median advertised rate | Highest advertised rate | Main current fields |
|---|---|---|---|---|---|
| Mercor | 430 | 34.3% | $80/hr | $400/hr | Languages, Generalist, Sciences |
| micro1 | 298 | 23.7% | $70/hr | $400/hr | Sciences, Languages, STEM and Coding |
| Turing | 221 | 17.6% | Hourly & per-task mix | — | STEM and Coding, Generalist, Sciences |
| AfterQuery | 206 | 16.4% | $75/hr | $200/hr | Languages, Medical, STEM and Coding |
| Ethos | 100 | 8% | $80/hr | $150/hr | Generalist, Finance, STEM and Coding |
| Handshake AI | Profile entry | — | Varies by matched opportunity | — | — |
Turing is marked “Hourly & per-task mix”: Sidequest publishes no hourly median for it here or on the platforms page, because one would misdescribe listings that pay per task; see each listing for the rate it states. For a side-by-side, see the platforms comparison, Mercor alternatives and micro1 alternatives.
Advertised pay for AI training work
Every rate on this page is a figure a listing advertises. It is not realised earnings: project availability, screening, eligibility and task volume all vary, and a listed band says what a platform offers an evaluator or expert for a role, not what a given person is paid over a month. Listings that pay per task or per project, or that publish no figure, are excluded from the hourly statistics rather than converted into one.
Across current Sidequest listings, 1007 advertise an hourly band with a positive floor. Each of those enters the statistics as the midpoint of its band, and the quartiles below are read from that list of midpoints — listing by listing, never by averaging the per-field or per-platform medians.
- Listings in rate calculations
- 1007
- 25th percentile
- $50/hr
- Median
- $78/hr
- 75th percentile
- $100/hr
- Highest advertised
- $400/hr
Advertised hourly rates by field
Median and highest advertised rate appear once a field has 3 hourly listings; the 25th and 75th percentiles need 5. Below those floors there is a number but not a statistic, and the row says so.
| Field | Median | 25th percentile | 75th percentile | Highest advertised | Hourly listings |
|---|---|---|---|---|---|
| Languages | $30 | $20 | $50 | $300 | 181 |
| Generalist | $80 | $65 | $100 | $280 | 159 |
| STEM and Coding | $80 | $70 | $100 | $280 | 107 |
| Sciences | $80 | $60 | $100 | $280 | 107 |
| Other | $78 | $65 | $100 | $250 | 88 |
| Finance | $100 | $80 | $100 | $300 | 98 |
| Medical | $88 | $75 | $120 | $400 | 96 |
| Legal | $115 | $100 | $150 | $400 | 87 |
| Creative | $55 | $38 | $75 | $300 | 69 |
| Education | $60 | $54 | $80 | $120 | 15 |
10 of 10 fields clear the floor for a median right now.
Advertised hourly rates by platform
Every platform on the board, with a figure only where the hourly sample is comparable. A platform whose listings mix hourly and per-task pay, or whose board presence is a profile entry, shows the reason instead of a figure — the same rule the platforms page applies.
| Platform | Median | 25th percentile | 75th percentile | Highest advertised | Hourly listings |
|---|---|---|---|---|---|
| Mercor | $80 | $55 | $100 | $400 | 386 |
| micro1 | $70 | $45 | $105 | $400 | 294 |
| Turing | Hourly & per-task mix | — | — | — | — |
| AfterQuery | $75 | $45 | $100 | $200 | 204 |
| Ethos | $80 | $70 | $100 | $150 | 100 |
| Handshake AI | Varies by matched opportunity | — | — | — | 0 |
4 of 6 platforms have a comparable hourly median right now.
Where AI training platforms are hiring
Languages, Generalist and STEM and Coding currently account for 46.3% of the listings tracked by Sidequest — Languages 202 (16.1%), Generalist 199 (15.9%) and STEM and Coding 180 (14.3%). This is the current inventory, not a fixed structure: a platform's next project wave can move a field by dozens of listings.
- Languages: Mercor 79 · AfterQuery 50 · micro1 44
- Generalist: Mercor 59 · Ethos 46 · Turing 37
- STEM and Coding: Turing 63 · Mercor 54 · micro1 36
Field pages break these fields down by specialty: STEM and Coding, Finance, Medical and Legal.
What this report does not measure
- Sidequest does not capture every AI training opportunity globally. The dataset is the platforms it tracks and the listings they publish.
- Advertised rates are not guaranteed earnings. A band is what a listing offers; screening, task volume and availability decide what anyone is paid.
- Private, invite-only or unlisted projects do not appear. A platform that matches from a profile may have work that is never published as a listing.
- Availability changes quickly. A field or platform can gain or lose dozens of listings between two visits, and the figures here move with it.
- Geography and eligibility vary by listing. A count says nothing about who may apply; each listing states its own restrictions.
- Non-hourly compensation is excluded from hourly statistics. Per-task and per-project listings are counted as listings but never converted into an hourly figure.
- No history is shown. Sidequest records a daily board snapshot, but this report describes the board as it is when rendered; trends are not inferred from listing dates.
Methodology
- Data source
- The current Sidequest role inventory — listings scraped or entered from the platforms it tracks — and the Sidequest platform registry for platform names and rate-display rules.
- What counts as a listing
- Active rows whose kind is a listing. A platform's profile entry (its “create a profile” door) and an open application pool are on the board but are not vacancies, and are excluded from every count and share here.
- Rates
- Exactly the rates page's method. A listing enters the rate statistics only if it advertises an hourly band with a positive floor; it enters as the midpoint of that band. The median is the middle of those midpoints. The 25th and 75th percentiles are linear-interpolation quantiles of the same midpoints (the method R, NumPy and spreadsheets use by default), which gives the same median. “Highest advertised” is the top of the highest band, a ceiling one listing states. Per-task and per-project listings, and listings with no figure, are excluded rather than estimated. A listing that advertises a $0 floor is left out of every statistic — a zero floor is not a low rate, it is the absence of one; 1 such listing is excluded right now.
- Floors
- A median or highest rate is published once a row has 3 hourly listings — the site's existing standard, so this page and the rates page publish the same rows. Percentiles need 5: with fewer, a quartile is an interpolation between two values, which is a number but not a statistic. A platform whose registry entry hides its hourly median because its listings mix hourly and per-task pay shows no median here either.
- Fields
- The current Sidequest classification taxonomy. Each listing is counted once, under the field Sidequest classified it in; the retired engineering category is folded into STEM and Coding. Field rows therefore sum to the total.
- Freshness
- Figures are generated from the current Sidequest inventory when this page is rendered — this render on 16 September 2026. The article itself was published 2026-09-16; no figure on it is stored or hand-typed.