Key findings

1
Wednesday and Thursday are the busiest posting days, with 52.1% of new postings between them (an even spread would give two days 28.6%); 90.9% carry a Monday-to-Friday date.
Figure 1
2
9.1% 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 24.2% of apply clicks.
Figure 1
3
Postings with a stated time are busiest in the 17:00 UTC hour (14.0% of them), and 36.2% fall in the 16:00 to 20:00 UTC block.
Figure 2
4
Apply clicks peak in the 14:00 UTC hour (6.9% of clicks), 3 hours before the busiest posting hour.
Figure 6
5
Wednesday is the busiest day both for applying (19.8% of an average week's apply clicks) and for posting (28.4% of new postings).
Figure 5
6
The busiest four hours of the week for applying are Saturday 16:00 to 20:00 UTC, with 5.6% of the week's apply clicks (an even spread would give each block 2.4%).
Figure 7
7
Half of the postings with a stated time were in their platform's public feed and read by this site within 12.1 hours, and 70.6% within a day; TransPerfect DataForce is the slowest (median 70.2 hours).
Figure 4
8
Mercor posts the most on weekends (17.8% of its dated postings); Mindrift and TransPerfect DataForce 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)	New postings	Share of an average week
Monday	41	8.1%
Tuesday	83	20.5%
Wednesday	115	28.4%
Thursday	96	23.7%
Friday	41	10.1%
Saturday	46	9.1%
Sunday	0	0.0%

Sample. 422 new listings dated from 1 Aug 2026 to 31 Aug 2026 by their job board, on 7 platforms; 938 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): AfterQuery and Mindrift. They count by weekday and are left out of the hours.

Left out: Braintrust and xAI (SpaceXAI), 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 (76 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: micro1 and Alignerr, whose boards give no posting date (269 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, August 2026 edition.

Figure 2

New postings by hour of the day

At what time of day are AI training roles posted?

Hour (UTC)	Share of postings	New postings
00:00 to 00:59	1.9%	5
01:00 to 01:59	2.6%	7
02:00 to 02:59	0.8%	2
03:00 to 03:59	0.4%	1
04:00 to 04:59	7.9%	21
05:00 to 05:59	4.9%	13
06:00 to 06:59	3.0%	8
07:00 to 07:59	2.3%	6
08:00 to 08:59	0.0%	0
09:00 to 09:59	2.3%	6
10:00 to 10:59	4.5%	12
11:00 to 11:59	0.4%	1
12:00 to 12:59	3.0%	8
13:00 to 13:59	1.9%	5
14:00 to 14:59	0.4%	1
15:00 to 15:59	1.9%	5
16:00 to 16:59	3.0%	8
17:00 to 17:59	14.0%	37
18:00 to 18:59	7.2%	19
19:00 to 19:59	12.1%	32
20:00 to 20:59	7.5%	20
21:00 to 21:59	8.7%	23
22:00 to 22:59	3.4%	9
23:00 to 23:59	6.0%	16

Sample. 265 new listings from 1 Aug 2026 to 31 Aug 2026 whose job board gives a posting date and time, on 5 platforms.

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

Mercor accounts for 78.5% 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: 40.0% 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): AfterQuery and Mindrift. They count by weekday and are left out of the hours.

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: micro1 and Alignerr, whose boards give no posting date (269 new listings).

Source. AITraining.jobs Research, When AI Training Roles Are Posted, August 2026 edition.

Figure 3

When each platform posts

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

Platform	New listings	Dated postings	Posted on a weekend	Busiest day	Busiest hour	Posting stamp
Mercor	352	208	17.8%	Wednesday	17:00 UTC	Date and time
AfterQuery	92	92	0.9%	Thursday	n/a	Date only
Mindrift	67	65	0.0%	Monday	n/a	Date only
Terac	32	27	3.1%	Wednesday	00:00 UTC	Date and time
TransPerfect DataForce	32	17	0.0%	Wednesday	07:00 UTC	Date and time
AppenSmall sample	10	9	0.0%	Wednesday	05:00 UTC	Date and time
Outlier (Scale AI)Small sample	8	4	0.0%	Friday	22:00 UTC	Date and time
micro1Small sample	267	0	n/a	n/a	n/a	No posting date
BraintrustSmall sample	54	0	n/a	n/a	n/a	Later than first sighting (left out)
xAI (SpaceXAI)Small sample	22	0	n/a	n/a	n/a	Later than first sighting (left out)
AlignerrSmall sample	2	0	n/a	n/a	n/a	No posting date

Sample. 938 new listings from 1 Aug 2026 to 31 Aug 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, August 2026 edition.

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?

Platform	Postings measured	Median hours to first sighting	Seen within a day	Within a week
All platforms with posting times	265	12.1	70.6%	98.1%
TransPerfect DataForce	17	70.2	29.4%	88.2%
AppenSmall sample	9	12.6	100.0%	100.0%
Mercor	208	12.1	68.8%	98.6%
Outlier (Scale AI)Small sample	4	10.2	75.0%	100.0%
Terac	27	8.2	100.0%	100.0%

Sample. 265 new listings from 1 Aug 2026 to 31 Aug 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 (100.0% 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, August 2026 edition.

Figure 5

Apply clicks by day of the week

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

Day (UTC)	Apply clicks	Share of an average week	Share of new postings
Monday	196	18.2%	8.1%
Tuesday	87	10.1%	20.5%
Wednesday	171	19.8%	28.4%
Thursday	138	16.0%	23.7%
Friday	102	11.8%	10.1%
Saturday	120	11.1%	9.1%
Sunday	141	13.1%	0.0%

Sample. 955 apply clicks on AITraining.jobs from 1 Aug 2026 to 31 Aug 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, August 2026 edition.

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	Apply clicks	Share of new postings
00:00 to 00:59Partial week	1.5%	14	1.9%
01:00 to 01:59	4.2%	40	2.6%
02:00 to 02:59	3.8%	36	0.8%
03:00 to 03:59Partial week	2.8%	27	0.4%
04:00 to 04:59Partial week	2.7%	26	7.9%
05:00 to 05:59Partial week	1.8%	17	4.9%
06:00 to 06:59Partial week	1.8%	17	3.0%
07:00 to 07:59Partial week	2.3%	22	2.3%
08:00 to 08:59	3.7%	35	0.0%
09:00 to 09:59	4.1%	39	2.3%
10:00 to 10:59	5.8%	55	4.5%
11:00 to 11:59	4.4%	42	0.4%
12:00 to 12:59	5.1%	49	3.0%
13:00 to 13:59	3.4%	32	1.9%
14:00 to 14:59	6.9%	66	0.4%
15:00 to 15:59	4.6%	44	1.9%
16:00 to 16:59	6.7%	64	3.0%
17:00 to 17:59	6.6%	63	14.0%
18:00 to 18:59	6.9%	66	7.2%
19:00 to 19:59	3.9%	37	12.1%
20:00 to 20:59	4.0%	38	7.5%
21:00 to 21:59	4.7%	45	8.7%
22:00 to 22:59	5.2%	50	3.4%
23:00 to 23:59	3.2%	31	6.0%

Sample. 955 apply clicks on AITraining.jobs from 1 Aug 2026 to 31 Aug 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.

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, August 2026 edition.

Figure 7

The week of applying, in four-hour blocks

Which hours of which days draw the most applications?

Day (UTC)	00:00 to 04:00	04:00 to 08:00	08:00 to 12:00	12:00 to 16:00	16:00 to 20:00	20:00 to 00:00	Whole day
Monday	1.0%	2.7%	4.1%	3.4%	5.2%	1.8%	18.2%
Tuesday	2.9%	1.0%	1.0%	1.7%	1.4%	2.0%	10.1%
Wednesday	2.8%	1.0%	4.7%	3.5%	3.8%	3.9%	19.8%
Thursday	2.1%	1.2%	2.4%	5.0%	3.1%	2.2%	16.0%
Friday	2.7%	0.8%	3.1%	1.0%	1.0%	3.1%	11.8%
Saturday	0.2%	0.3%	0.4%	2.2%	5.6%	2.5%	11.1%
Sunday	1.3%	1.4%	2.3%	3.1%	3.1%	1.9%	13.1%

Sample. 955 apply clicks on AITraining.jobs from 1 Aug 2026 to 31 Aug 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.

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, August 2026 edition.

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.

Live dataLast 28 complete days, recomputed daily
September 2026 editionFrozen 10 Oct 2026
August 2026 editionFrozen 10 Oct 2026

Cite and reuse

Free to quote, chart and republish under CC BY 4.0 with a link to this page. This edition is frozen, so its figures will read the same later.

Citation

AITraining.jobs (2026). When AI Training Roles Are Posted, August 2026 edition. https://www.aitraining.team/research/when-ai-training-roles-are-posted/2026-08

BibTeX

@misc{aitrainingjobs_when_ai_training_roles_are_posted_2026_08,
  author = {{AITraining.jobs}},
  title = {When AI Training Roles Are Posted, August 2026 edition},
  year = {2026},
  howpublished = {\url{https://www.aitraining.team/research/when-ai-training-roles-are-posted/2026-08}},
}

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