Skip to content

Research report ยท Live data

When AI Training Roles Are Posted

From 20 Sep 2026 to 17 Oct 2026, 92.4% of dated AI training postings fell on a weekday (Wednesday the busiest day), while people applied most on Thursday and in the 17:00 UTC hour.

Data window
20 Sep 2026 to 17 Oct 2026
28 days, UTC
Computed
18 Oct 2026, 15:17 UTC
Recomputed every day
Source
The posting dates and times the AI training platforms' public job boards give for new listings, read every day, and the apply clicks made on AITraining.jobs, counted by weekday and hour only.
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.

of new postings dated on a weekend
8%
busiest posting day (40.8% of new postings)
Wednesday
busiest hour for apply clicks
17:00 UTC
median from posting to the public feed
15 hours

Key findings

  1. 1

    Wednesday and Thursday are the busiest posting days, with 56.0% of new postings between them (an even spread would give two days 28.6%); 92.4% carry a Monday-to-Friday date.

    Figure 1
  2. 2

    7.6% of new postings are dated on a Saturday or Sunday (UTC), against 28.6% if postings were spread evenly over the week, while weekends draw 25.4% of apply clicks.

    Figure 1
  3. 3

    Postings with a stated time are busiest in the 10:00 UTC hour (15.1% of them), and 27.8% fall in the 08:00 to 12:00 UTC block.

    Figure 2
  4. 4

    Apply clicks peak in the 17:00 UTC hour (6.1% of clicks), 7 hours after the busiest posting hour.

    Figure 6
  5. 5

    Thursday is the busiest day for applying (17.4% of an average week's apply clicks), while Wednesday is the busiest for posting (40.8%).

    Figure 5
  6. 6

    The busiest four hours of the week for applying are Wednesday 16:00 to 20:00 UTC, with 3.7% of the week's apply clicks (an even spread would give each block 2.4%).

    Figure 7
  7. 7

    Half of the postings with a stated time were in their platform's public feed and read by this site within 15.1 hours, and 76.7% within a day; Handshake AI is the slowest (median 254.3 hours).

    Figure 4
  8. 8

    Appen posts the most on weekends (33.4% of its dated postings); 6 of the 11 platforms with at least 10 dated postings dated none of theirs on a weekend.

    Figure 3

Figure 1

New postings by day of the week

On which days of the week are AI training roles posted?

Day (UTC)Share of an average week
Monday
10.2%
Tuesday
14.3%
Wednesday
40.8%
Thursday
15.2%
Friday
12.0%
Saturday
6.3%
Sunday
1.3%

Sample. 1,035 new listings dated from 20 Sep 2026 to 17 Oct 2026 by their job board, on 17 platforms; 2,252 new listings in all.

The day is the posting date the job board gives, in UTC. A share is the day's postings per day of that weekday in the window, so a month with five of some weekdays is not tilted by it.

Date-only stamps (at least 90% of a platform's stamps at exactly 00:00:00 UTC): Mindrift and AfterQuery. They count by weekday and are left out of the hours.

Left out: Braintrust, whose stamps are often later than the day this site first saw the role (at least 25% more than an hour after it), so they record an edit or re-publication, not the posting (32 new listings).

No platform's stamps are this site's own read time: on every platform fewer than half fall within 10 minutes of the first sighting.

Left out: Alignerr and micro1, whose boards give no posting date (690 new listings).

Each platform's first week in this dataset (its back catalogue) is not counted as new.

Source. AITraining.jobs Research, When AI Training Roles Are Posted, live data.

Figure 2

New postings by hour of the day

At what time of day are AI training roles posted?

Hour (UTC)Share of postings
00:00 to 00:59
3.4%
01:00 to 01:59
3.1%
02:00 to 02:59
3.1%
03:00 to 03:59
1.2%
04:00 to 04:59
2.7%
05:00 to 05:59
2.7%
06:00 to 06:59
14.8%
07:00 to 07:59
2.7%
08:00 to 08:59
3.6%
09:00 to 09:59
3.7%
10:00 to 10:59
15.1%
11:00 to 11:59
5.4%
12:00 to 12:59
3.1%
13:00 to 13:59
3.9%
14:00 to 14:59
3.7%
15:00 to 15:59
2.2%
16:00 to 16:59
2.4%
17:00 to 17:59
3.1%
18:00 to 18:59
4.6%
19:00 to 19:59
0.7%
20:00 to 20:59
3.9%
21:00 to 21:59
5.1%
22:00 to 22:59
3.4%
23:00 to 23:59
2.5%

Sample. 589 new listings from 20 Sep 2026 to 17 Oct 2026 whose job board gives a posting date and time, on 15 platforms.

The hour of the posting time the job board gives, in UTC.

OpenTrain AI accounts for 49.9% of these postings, so its schedule shapes the curve; each platform's busiest hour is in the platform table.

Roles are often posted in batches: 38.9% of these postings carry a stamp within the same 10 minutes as at least 4 others on the same platform, so a few batches can make one hour stand out.

Date-only stamps (at least 90% of a platform's stamps at exactly 00:00:00 UTC): Mindrift and AfterQuery. They count by weekday and are left out of the hours.

Left out: Braintrust, whose stamps are often later than the day this site first saw the role (at least 25% more than an hour after it), so they record an edit or re-publication, not the posting (32 new listings).

No platform's stamps are this site's own read time: on every platform fewer than half fall within 10 minutes of the first sighting.

Left out: Alignerr and micro1, whose boards give no posting date (690 new listings).

Source. AITraining.jobs Research, When AI Training Roles Are Posted, live data.

Figure 3

When each platform posts

Which platforms post on weekends, and on which day and hour does each post most?

PlatformDated postingsPosted on a weekendBusiest day
Mindrift429
0.0%
Wednesday
OpenTrain AI294
19.4%
Wednesday
Mercor108
8.4%
Wednesday
Welo Data39
0.0%
Friday
Appen30
33.4%
Wednesday
Turing30
0.0%
Monday
Handshake AI22
4.5%
Monday
Sigma AI22
0.0%
Thursday
TransPerfect DataForce17
0.0%
Tuesday
AfterQuery17
0.0%
Friday
Terac14
7.1%
Thursday
Anyone AISmall sample6
0.0%
Monday
xAI (SpaceXAI)Small sample3
0.0%
Friday
InvisibleSmall sample1
0.0%
Tuesday
Perle LabsSmall sample1
0.0%
Monday
Sovrano AISmall sample1
0.0%
Friday
SuperAnnotate (SME)Small sample1
0.0%
Wednesday
AlignerrSmall sample0
n/a
n/a
micro1Small sample0
n/a
n/a
BraintrustSmall sample0
n/a
n/a
Outlier (Scale AI)Small sample0
n/a
n/a

Sample. 2,252 new listings from 20 Sep 2026 to 17 Oct 2026, by the platform that lists them. Listings only.

Dated postings: new listings whose board gives a usable posting date inside the window. Weekend: Saturday and Sunday in UTC, as a share of the platform's average week (an even spread would give 28.6%).

Busiest hour reads n/a for a board that gives a date without a time.

Platforms with fewer than 10 dated postings are marked as a small sample.

Source. AITraining.jobs Research, When AI Training Roles Are Posted, live data.

Figure 4

From posting to the public feed

How soon after a job board's posting time does a role appear in its public feed?

PlatformMedian hours to first sightingSeen within a day
All platforms with posting times
15.1
76.7%
Handshake AI
254.3
36.4%
Perle LabsSmall sample
232.1
0.0%
Sovrano AISmall sample
143.6
0.0%
Turing
28.7
50.0%
Appen
23.1
56.7%
Welo Data
19.3
66.7%
TransPerfect DataForce
17.8
76.5%
SuperAnnotate (SME)Small sample
17.0
100.0%
Sigma AI
15.0
95.5%
OpenTrain AI
14.0
81.0%
InvisibleSmall sample
13.2
100.0%
Terac
8.4
100.0%
Mercor
8.0
82.4%
xAI (SpaceXAI)Small sample
7.8
100.0%
Anyone AISmall sample
3.9
100.0%

Sample. 589 new listings from 20 Sep 2026 to 17 Oct 2026 whose board gives a posting date and time inside the window.

Hours from the board's posting time to the first time this site read the role in the platform's public feed. Feeds are read about once a day (57.8% of new listings were first seen in the 06:00 UTC hour), so up to a day of each lag is the wait for the next read; a lag of several days means the role reached the public feed well after the date its board gives.

Boards that give only a date are left out: their lag would run from midnight of the stated day.

Only postings dated inside the window are measured, so a lag longer than the window cannot appear.

Platforms with fewer than 10 postings measured are marked as a small sample.

Source. AITraining.jobs Research, When AI Training Roles Are Posted, live data.

Figure 5

Apply clicks by day of the week

On which days of the week do people apply for AI training work?

Day (UTC)Share of an average week
Monday
15.2%
Tuesday
14.6%
Wednesday
15.1%
Thursday
17.4%
Friday
12.4%
Saturday
14.2%
Sunday
11.2%

Sample. 6,200 apply clicks on AITraining.jobs from 20 Sep 2026 to 17 Oct 2026, by the weekday and hour they were made in UTC.

Human apply clicks on AITraining.jobs, on specific roles and on platforms' general sign-up links. Counted site-wide, never by platform.

Clicks follow where AITraining.jobs visitors live and when they browse: every hour and day here is UTC (09:00 UTC is early morning in the Americas and evening in East Asia), so a peak shows when this audience is awake, not when the market as a whole applies.

The last column repeats the share of new postings from the posting figure, for comparison.

Source. AITraining.jobs Research, When AI Training Roles Are Posted, live data.

Figure 6

Apply clicks by hour of the day

At what time of day do people apply for AI training work?

Hour (UTC)Share of apply clicks
00:00 to 00:59
3.4%
01:00 to 01:59
2.9%
02:00 to 02:59
3.1%
03:00 to 03:59
3.2%
04:00 to 04:59
2.3%
05:00 to 05:59
3.7%
06:00 to 06:59
4.6%
07:00 to 07:59
5.1%
08:00 to 08:59
3.7%
09:00 to 09:59
4.2%
10:00 to 10:59
3.8%
11:00 to 11:59
4.3%
12:00 to 12:59
4.3%
13:00 to 13:59
3.7%
14:00 to 14:59
5.2%
15:00 to 15:59
5.4%
16:00 to 16:59
4.7%
17:00 to 17:59
6.1%
18:00 to 18:59
4.6%
19:00 to 19:59
5.4%
20:00 to 20:59
4.7%
21:00 to 21:59
3.0%
22:00 to 22:59
4.2%
23:00 to 23:59
4.3%

Sample. 6,200 apply clicks on AITraining.jobs from 20 Sep 2026 to 17 Oct 2026, by the weekday and hour they were made in UTC.

Human apply clicks on AITraining.jobs, on specific roles and on platforms' general sign-up links. Counted site-wide, never by platform.

Clicks follow where AITraining.jobs visitors live and when they browse: every hour and day here is UTC (09:00 UTC is early morning in the Americas and evening in East Asia), so a peak shows when this audience is awake, not when the market as a whole applies.

The last column repeats the share of new postings with a stated time in that hour, for comparison.

Source. AITraining.jobs Research, When AI Training Roles Are Posted, live data.

Figure 7

The week of applying, in four-hour blocks

Which hours of which days draw the most applications?

Day (UTC)00:00 to 04:0004:00 to 08:0008:00 to 12:0012:00 to 16:0016:00 to 20:0020:00 to 00:00
Monday2.2%2.3%2.9%2.6%2.5%2.7%
Tuesday1.2%2.1%2.3%3.4%3.6%2.0%
Wednesday1.6%2.1%2.0%3.3%3.7%2.4%
Thursday2.3%3.2%3.5%2.2%2.9%3.3%
Friday1.7%2.9%1.6%1.9%2.6%1.7%
Saturday2.0%1.8%2.0%2.9%3.1%2.3%
Sunday1.5%1.5%1.6%2.3%2.3%2.0%

Sample. 6,200 apply clicks on AITraining.jobs from 20 Sep 2026 to 17 Oct 2026, by the weekday and hour they were made in UTC.

Human apply clicks on AITraining.jobs, on specific roles and on platforms' general sign-up links. Counted site-wide, never by platform.

Clicks follow where AITraining.jobs visitors live and when they browse: every hour and day here is UTC (09:00 UTC is early morning in the Americas and evening in East Asia), so a peak shows when this audience is awake, not when the market as a whole applies.

Each cell is that block's share of an average week's apply clicks (per day of that weekday in the window); all cells add up to 100%.

Source. AITraining.jobs Research, When AI Training Roles Are Posted, 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 posting dates and times the AI training platforms' public job boards give for new listings, read every day, and the apply clicks made on AITraining.jobs, counted by weekday and hour only. 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.

Posting time
The date and time a job board gives for a new listing, in UTC. Only postings dated inside the window are counted.
New listing
A role that first appeared during the window on a platform already tracked before that week. A platform's first week in this dataset is its back catalogue and is not counted as new.
Posting stamp
What a board's posting field holds, read from its own new listings: a date and time; a date only (at least 90% at exactly midnight UTC, counted by weekday but not by hour); this site's own read time (at least half within 10 minutes of the first sighting, left out); a time later than this site's first sighting (at least a quarter more than an hour after it, an edit time, left out); or no date (left out).
Share of an average week
A weekday's count per day of that weekday in the window, over the same rate summed across the week, so a month with five of some weekdays is not tilted by it.
Weekend
Saturday and Sunday in UTC. An even spread over the week would put 28.6% there.
Lag to the feed
Hours from a board's posting time to the first time this site read the role in its public feed. Feeds are read about once a day, so up to a day of the lag is the wait for the next read.
Apply click
A click on an apply button on AITraining.jobs that opens a platform's application page, on a specific role or on a platform's general sign-up link. Made by a person: automated traffic is set aside as it is recorded. Counted site-wide, never by platform.
Window
Live: the last 28 complete days, in UTC. Edition: one calendar month, in UTC.

What this data cannot tell you

  • Apply clicks follow where AITraining.jobs visitors live and when they browse. Every day and hour is stated in UTC, so a peak shows when this audience is awake, not when the market as a whole applies.
  • Posting times are what the job boards state. A date-only board may give its local date, and boards that give no date, or a date later than the role first appeared, are left out (each figure names them).
  • The largest boards dominate the hour figures, and roles are often posted in batches, so a few batches can make one hour stand out.
  • Platforms that recruit without a public job board are not counted.
  • An apply click starts an application. It is not a submitted application or a hire, and what is shown first gets clicked more.

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.

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). When AI Training Roles Are Posted (live data, 20 Sep 2026 to 17 Oct 2026). https://www.aitraining.team/research/when-ai-training-roles-are-posted (accessed 18 Oct 2026)

BibTeX

@misc{aitrainingjobs_when_ai_training_roles_are_posted_live,
  author = {{AITraining.jobs}},
  title = {When AI Training Roles Are Posted (live data, 20 Sep 2026 to 17 Oct 2026)},
  year = {2026},
  howpublished = {\url{https://www.aitraining.team/research/when-ai-training-roles-are-posted}},
  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