data labeling vendors

ranked on what they'll tell you: what it costs, who actually does the labelling, and what the public record says about how those people are treated.

5 listed · 5 researched by us · featured first, then most upvoted

Snorkel AI

we checked thisnot published

reduces how much human labelling you need in the first place, which is the only structural answer to this category's problems.

#teams-with-domain-heavy-#documents-who-would-rath

Argilla

we checked thisfree

free open-source annotation tooling with no workforce attached — which is both its strength and the reason it can't finish first.

#teams-with-their-own-ann#or-open-communities-labe

Labelbox

we checked thisnot published

annotation platform plus an expert marketplace, with the highest reported pay band of the crowd-model vendors.

#teams-wanting-tooling#on-demand-domain-experts#including-medical

Toloka

we checked thisnot published

the largest open crowd here at the lowest reported pay, and a corporate lineage worth tracing before you sign.

#very-high-volume-microta

Surge AI

we checked thisnot published

bootstrapped, profitable and serving the frontier labs — facing a class action over how it classifies the people doing the work.

#frontier-lab-grade-rlhf#reasoning-data-where-qua

this page is deliberately short: a handful of data labeling vendors we looked at properly, rather than everything that exists. if you build one that belongs here, a free listing takes two minutes and $50 pins it to the top of this page.

add a tool →