Key findings
- 1
English (163), Spanish (109), German (107), Arabic (100) and French (80) had the most open roles on 17 Oct 2026, of 95 languages named in role titles.
- 2
Among languages with at least 10 open roles at the start, Polish grew fastest (+113.3%, 15 to 32) and Thai shrank the most (-56%).
- 3
Arabic roles advertised the highest median pay, $100/hr, and Telugu the lowest, $20/hr, among languages with at least 8 priced roles.
- 4
Per open role, English roles drew the most applicant interest (1.71x the average role on the site) and German roles the least (0.35x).
- 5
93.3% of the 1,004 open multilingual roles name the language they need; together, open roles named 95 languages.
Figure 1
Open roles by language
Which languages are AI training platforms hiring for?
| Language | Open on 17 Oct 2026 | New listings |
|---|---|---|
| English | 163 | 92 |
| Spanish | 109 | 60 |
| German | 107 | 73 |
| Arabic | 100 | 60 |
| French | 80 | 46 |
| Portuguese | 75 | 52 |
| Chinese (Mandarin) | 69 | 32 |
| Korean | 69 | 49 |
| Japanese | 67 | 52 |
| Italian | 61 | 36 |
| Dutch | 54 | 27 |
| Hebrew | 35 | 26 |
| Polish | 35 | 38 |
| Swedish | 34 | 39 |
| Danish | 29 | 21 |
| Hindi | 29 | 11 |
| Finnish | 28 | 16 |
| Vietnamese | 27 | 7 |
| Indonesian | 24 | 17 |
| Russian | 22 | 8 |
| Bengali | 21 | 11 |
| Telugu | 20 | 3 |
| Marathi | 18 | 10 |
| Tamil | 18 | 7 |
| Thai | 18 | 4 |
| Czech | 15 | 3 |
| Norwegian | 15 | 4 |
| Turkish | 15 | 7 |
| Filipino (Tagalog)Small sample | 14 | 8 |
| Punjabi | 14 | 5 |
| Gujarati | 12 | 7 |
| Hungarian | 12 | 3 |
| MalaySmall sample | 12 | 9 |
| Croatian | 11 | 0 |
| Romanian | 11 | 2 |
| CatalanSmall sample | 10 | 7 |
| Ukrainian | 10 | 1 |
| Urdu | 10 | 4 |
| BulgarianSmall sample | 9 | 1 |
| CantoneseSmall sample | 9 | 5 |
Sample. 1,616 of the 3,668 roles open at the end of 17 Oct 2026 name a language in their title; 95 languages are named. New listings from 20 Sep 2026 to 17 Oct 2026.
A role counts for a language when its title names it (for example 'Korean Language Specialist' or 'Bilingual Spanish Physician'). A role naming two languages counts for both. English counts only where the title names it.
Change compares open roles at the start and the end of the window on the platforms tracked throughout; it is marked as a small sample under 10 roles at the start.
Source. AITraining.jobs Research, Languages in Demand for AI Training, live data.
Figure 2
Advertised hourly pay by language
Which languages pay the most?
| Language | Median per hour | Middle half |
|---|---|---|
| Arabic | $100 | $55 to $100 |
| Danish | $100 | $43 to $100 |
| Finnish | $100 | $36 to $100 |
| Hebrew | $100 | $76 to $100 |
| Polish | $100 | $39 to $100 |
| Dutch | $65 | $26 to $100 |
| German | $65 | $34 to $100 |
| Italian | $65 | $26 to $100 |
| Japanese | $65 | $31 to $100 |
| Korean | $65 | $31 to $100 |
| Portuguese | $65 | $30 to $100 |
| Swedish | $65 | $36 to $100 |
| Norwegian | $55 | $37 to $60 |
| Spanish | $51 | $25 to $65 |
| French | $50 | $25 to $65 |
| Filipino (Tagalog) | $37 | $28 to $73 |
| Hindi | $37 | $23 to $65 |
| Indonesian | $37 | $21 to $65 |
| Russian | $37 | $30 to $65 |
| Chinese (Mandarin) | $35 | $23 to $45 |
| Hungarian | $33 | $26 to $36 |
| Czech | $30 | $21 to $37 |
| English | $30 | $30 to $65 |
| Marathi | $30 | $20 to $65 |
| Thai | $30 | $15 to $37 |
| Turkish | $29 | $23 to $37 |
| Malay | $28 | $21 to $36 |
| Vietnamese | $28 | $16 to $65 |
| Croatian | $27 | $15 to $30 |
| Catalan | $26 | $22 to $36 |
| Persian (Farsi) | $25 | $6 to $31 |
| Punjabi | $25 | $14 to $49 |
| Tamil | $24 | $16 to $33 |
| Bengali | $23 | $9 to $48 |
| Malayalam | $23 | $16 to $30 |
| Gujarati | $20 | $10 to $30 |
| Telugu | $20 | $14 to $33 |
Sample. Roles open on 17 Oct 2026 that state an hourly rate; languages with at least 8 such roles.
Pay is the hourly figure the posting advertises (the middle of a stated range). The middle half runs from the 25th to the 75th percentile.
Source. AITraining.jobs Research, Languages in Demand for AI Training, live data.
Figure 3
Applicant demand per open role, by language
Which languages draw the most applicants per opening, and which are short of them?
| Language | Clicks per role a week | Index |
|---|---|---|
| SwahiliSmall sample | 2.17 | 5.75x |
| KhmerSmall sample | 1.08 | 2.87x |
| KurdishSmall sample | 0.90 | 2.39x |
| HausaSmall sample | 0.75 | 1.99x |
| IgboSmall sample | 0.75 | 1.99x |
| English | 0.65 | 1.71x |
| Vietnamese | 0.60 | 1.58x |
| AmharicSmall sample | 0.57 | 1.50x |
| UrduSmall sample | 0.52 | 1.38x |
| Hindi | 0.50 | 1.32x |
| Indonesian | 0.50 | 1.32x |
| Chinese (Mandarin) | 0.49 | 1.28x |
| BelarusianSmall sample | 0.44 | 1.17x |
| Spanish | 0.44 | 1.16x |
| Japanese | 0.42 | 1.10x |
| Portuguese | 0.41 | 1.08x |
| AfrikaansSmall sample | 0.38 | 1.00x |
| BengaliSmall sample | 0.38 | 1.01x |
| YorubaSmall sample | 0.38 | 0.99x |
| Russian | 0.37 | 0.99x |
| TamilSmall sample | 0.35 | 0.93x |
| TurkishSmall sample | 0.35 | 0.91x |
| HebrewSmall sample | 0.33 | 0.87x |
| MacedonianSmall sample | 0.33 | 0.88x |
| Arabic | 0.28 | 0.73x |
| TeluguSmall sample | 0.26 | 0.68x |
| AssameseSmall sample | 0.21 | 0.55x |
| BulgarianSmall sample | 0.21 | 0.56x |
| GujaratiSmall sample | 0.21 | 0.57x |
| ItalianSmall sample | 0.19 | 0.49x |
| KazakhSmall sample | 0.19 | 0.50x |
| Korean | 0.19 | 0.51x |
| Filipino (Tagalog)Small sample | 0.16 | 0.42x |
| SerbianSmall sample | 0.16 | 0.44x |
| UkrainianSmall sample | 0.16 | 0.41x |
| GeorgianSmall sample | 0.15 | 0.41x |
| KannadaSmall sample | 0.15 | 0.39x |
| MarathiSmall sample | 0.15 | 0.39x |
| NepaliSmall sample | 0.15 | 0.39x |
| PunjabiSmall sample | 0.15 | 0.40x |
| CzechSmall sample | 0.14 | 0.36x |
| French | 0.14 | 0.37x |
| German | 0.13 | 0.35x |
| LatvianSmall sample | 0.13 | 0.35x |
| MalaySmall sample | 0.13 | 0.35x |
| PolishSmall sample | 0.13 | 0.35x |
| SwedishSmall sample | 0.13 | 0.35x |
| ThaiSmall sample | 0.13 | 0.35x |
| CantoneseSmall sample | 0.12 | 0.32x |
| MalayalamSmall sample | 0.12 | 0.33x |
| RomanianSmall sample | 0.12 | 0.31x |
| DutchSmall sample | 0.11 | 0.28x |
| NorwegianSmall sample | 0.11 | 0.30x |
| FinnishSmall sample | 0.09 | 0.23x |
| Haitian CreoleSmall sample | 0.08 | 0.20x |
| EstonianSmall sample | 0.07 | 0.18x |
| BasqueSmall sample | 0.06 | 0.16x |
| CroatianSmall sample | 0.06 | 0.17x |
| Persian (Farsi)Small sample | 0.06 | 0.15x |
| CatalanSmall sample | 0.05 | 0.14x |
| DanishSmall sample | 0.04 | 0.11x |
| IcelandicSmall sample | 0.04 | 0.10x |
| OdiaSmall sample | 0.04 | 0.10x |
| GreekSmall sample | 0.03 | 0.08x |
| SlovakSmall sample | 0.03 | 0.09x |
| SlovenianSmall sample | 0.03 | 0.08x |
| HungarianSmall sample | 0.02 | 0.06x |
| LithuanianSmall sample | 0.02 | 0.06x |
| AlbanianSmall sample | 0.00 | 0.00x |
| ArmenianSmall sample | 0.00 | 0.00x |
| AymaraSmall sample | 0.00 | 0.00x |
| AzerbaijaniSmall sample | 0.00 | 0.00x |
| BhojpuriSmall sample | 0.00 | 0.00x |
| BosnianSmall sample | 0.00 | 0.00x |
| BurmeseSmall sample | 0.00 | 0.00x |
| CebuanoSmall sample | 0.00 | 0.00x |
| GalicianSmall sample | 0.00 | 0.00x |
| Irish (Gaelic)Small sample | 0.00 | 0.00x |
| JavaneseSmall sample | 0.00 | 0.00x |
| KinyarwandaSmall sample | 0.00 | 0.00x |
| LaoSmall sample | 0.00 | 0.00x |
| LuxembourgishSmall sample | 0.00 | 0.00x |
| MalteseSmall sample | 0.00 | 0.00x |
| MฤoriSmall sample | 0.00 | 0.00x |
| MongolianSmall sample | 0.00 | 0.00x |
| NahuatlSmall sample | 0.00 | 0.00x |
| PapiamentoSmall sample | 0.00 | 0.00x |
| PashtoSmall sample | 0.00 | 0.00x |
| QuechuaSmall sample | 0.00 | 0.00x |
| SinhalaSmall sample | 0.00 | 0.00x |
| SomaliSmall sample | 0.00 | 0.00x |
| UzbekSmall sample | 0.00 | 0.00x |
| WelshSmall sample | 0.00 | 0.00x |
| WolofSmall sample | 0.00 | 0.00x |
| ZuluSmall sample | 0.00 | 0.00x |
Sample. Apply clicks on AITraining.jobs on roles naming each language, from 20 Sep 2026 to 17 Oct 2026, against the days those roles were open.
Clicks per role a week = apply clicks / (open role-days / 7). Index 1.00 is the average across every open role on the site.
Rows with fewer than 30 clicks or 3 roles open on an average day are marked as a small sample.
AITraining.jobs shows visitors the roles open in their country first, which affects roles limited to some countries.
Source. AITraining.jobs Research, Languages in Demand for AI Training, live data.
Figure 4
Who hires for each language
Is each language's hiring spread across platforms or held by one?
| Language | Largest platform | Its share | Platforms listing |
|---|---|---|---|
| English | AfterQuery | 28.8% | 20 |
| Spanish | Alignerr | 20.2% | 18 |
| German | Mindrift | 33.6% | 13 |
| Arabic | Mindrift | 38.0% | 14 |
| French | Invisible | 26.3% | 17 |
| Portuguese | Mindrift | 32.0% | 17 |
| Chinese (Mandarin) | Invisible | 15.9% | 16 |
| Korean | Mindrift | 24.6% | 17 |
| Japanese | Mindrift | 20.9% | 16 |
| Italian | Mindrift | 27.9% | 15 |
| Dutch | Mindrift | 31.5% | 12 |
| Hebrew | Mindrift | 54.3% | 10 |
| Polish | Mindrift | 57.1% | 11 |
| Swedish | Mindrift | 35.3% | 11 |
| Danish | Mindrift | 41.4% | 13 |
| Hindi | Invisible | 44.8% | 10 |
| Finnish | Mindrift | 42.9% | 12 |
| Vietnamese | Appen | 14.8% | 13 |
| Indonesian | Invisible | 29.2% | 13 |
| Russian | Invisible | 31.8% | 8 |
| Bengali | Sigma AI | 19.0% | 12 |
| Telugu | Invisible | 15.0% | 11 |
| Marathi | Invisible | 22.2% | 9 |
| Tamil | Sigma AI | 16.7% | 10 |
| Thai | Invisible | 16.7% | 10 |
| Czech | Welo Data | 26.7% | 8 |
| Norwegian | Invisible | 13.3% | 11 |
| Turkish | Welo Data | 26.7% | 8 |
| Filipino (Tagalog) | Invisible | 28.6% | 8 |
| Punjabi | Invisible | 21.4% | 7 |
| Gujarati | Braintrust | 16.7% | 8 |
| Hungarian | Invisible | 33.3% | 7 |
| Malay | Invisible | 25.0% | 9 |
| Croatian | Welo Data | 36.4% | 7 |
| Romanian | Sigma AI | 27.3% | 7 |
| Catalan | SuperAnnotate (SME) | 30.0% | 6 |
| Ukrainian | Sigma AI | 30.0% | 7 |
| Urdu | SuperAnnotate (SME) | 20.0% | 8 |
| BulgarianSmall sample | Invisible | 22.2% | 6 |
| CantoneseSmall sample | micro1 | 44.4% | 5 |
Sample. Roles open on 17 Oct 2026, by the platform that lists them.
A high share means one platform holds most of the openings: an applicant has few alternatives, and that platform sets the going rate.
Source. AITraining.jobs Research, Languages in Demand for AI Training, live data.
Figure 5
Where each language's roles are open
Can speakers apply from anywhere, or only from some countries?
| Language | Open worldwide | Only listed countries |
|---|---|---|
| CantoneseSmall sample | 44.4% | 55.6% |
| Vietnamese | 40.7% | 51.9% |
| Korean | 37.7% | 59.4% |
| Japanese | 37.3% | 56.7% |
| Filipino (Tagalog) | 35.7% | 57.1% |
| Hungarian | 33.3% | 58.3% |
| Norwegian | 33.3% | 60.0% |
| Thai | 33.3% | 50.0% |
| Turkish | 33.3% | 60.0% |
| Spanish | 31.2% | 64.2% |
| Bengali | 28.6% | 61.9% |
| Tamil | 27.8% | 66.7% |
| Hindi | 27.6% | 72.4% |
| Russian | 27.3% | 68.2% |
| German | 27.1% | 69.2% |
| Czech | 26.7% | 66.7% |
| Indonesian | 25.0% | 66.7% |
| Italian | 24.6% | 70.5% |
| Dutch | 24.1% | 72.2% |
| Chinese (Mandarin) | 23.2% | 71.0% |
| English | 22.7% | 41.7% |
| Portuguese | 22.7% | 72.0% |
| Swedish | 20.6% | 73.5% |
| Catalan | 20.0% | 80.0% |
| French | 20.0% | 73.8% |
| Telugu | 20.0% | 70.0% |
| Urdu | 20.0% | 70.0% |
| Romanian | 18.2% | 72.7% |
| Arabic | 17.0% | 82.0% |
| Gujarati | 16.7% | 66.7% |
| Malay | 16.7% | 75.0% |
| Marathi | 16.7% | 77.8% |
| Finnish | 14.3% | 78.6% |
| Polish | 14.3% | 82.9% |
| Punjabi | 14.3% | 78.6% |
| Danish | 13.8% | 79.3% |
| BulgarianSmall sample | 11.1% | 77.8% |
| Croatian | 9.1% | 81.8% |
| Hebrew | 5.7% | 88.6% |
| Ukrainian | 0.0% | 90.0% |
Sample. Roles open on 17 Oct 2026; shares of each language's own open roles.
The rule is read from each posting's location text and, where that names no country, from what the posting's own description states. The rest of each row is roles open worldwide except some countries, or with no rule stated.
Source. AITraining.jobs Research, Languages in Demand for AI Training, 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. Languages are read from role titles; apply clicks are made on AITraining.jobs. 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.
- Language of a role
- A role counts for a language when its title names it, for example 'Korean Language Specialist' or 'Bilingual Spanish Physician'. A role naming two languages counts for both. About nine in ten multilingual roles name their language.
- Open role
- A role listed by a given day and still in its platform's feed that day.
- 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.
- 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.
- Clicks per open role a week
- Apply clicks on AITraining.jobs divided by the days the roles were open, times 7.
- Demand index
- A group's clicks per open role a week divided by the same rate across every open role on the site. Above 1, its roles draw more applicants than the average role.
- Location rule
- Read from each posting's location text and, where that names no country, from what the posting's own description states: open worldwide, worldwide except listed countries, only listed countries, or no rule stated.
- Window
- Live: the last 28 complete days, in UTC. Edition: one calendar month, in UTC.
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 role that needs a language but does not name it in its title is not counted for it.
- Apply clicks measure interest among the people who use AITraining.jobs, 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. 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). Languages in Demand for AI Training (live data, 20 Sep 2026 to 17 Oct 2026). https://aitraining.run/research/languages-in-demand (accessed 18 Oct 2026)
BibTeX
@misc{aitrainingjobs_languages_in_demand_live,
author = {{AITraining.jobs}},
title = {Languages in Demand for AI Training (live data, 20 Sep 2026 to 17 Oct 2026)},
year = {2026},
howpublished = {\url{https://aitraining.run/research/languages-in-demand}},
note = {Accessed 18 Oct 2026},
}More research
- Expert Fields in Demand for AI TrainingWhich professions and disciplines are AI labs hiring to train models, and what do they pay?Read the report
- What AI Training Roles RequireWhat do AI training roles ask for: degrees, credentials, experience, hours, tests?Read the report
- AI Training Job TitlesWhat is AI training work called, and does the title say anything about the pay?Read the report
- Programming Languages in AI TrainingWhich programming languages are AI labs hiring coders to train models in, and what does each pay?Read the report
- What AI Trainers DoWhat tasks does AI training work involve: writing prompts, rating answers, red-teaming, coding?Read the report
- The Expert Scarcity Index for AI TrainingWhich expertise and languages are hardest to recruit for AI training work right now?Read the report
- Real-World Data Collection WorkWho is hiring people to collect real-world data for robots and physical AI, where, and at what pay?Read the report
- The State of AI Training PayWhat does AI training work pay, by kind of work and by platform?Read the report