Data annotation pricing
Most annotation vendors keep prices off their websites: 6 of the 18 we compared publish anything at all, and three put dollar amounts on the page. Label Your Data starts at two cents a box (or about $6 an annotator hour), Segments.ai works out at $2.67 an hour of platform use, and HumanSignal charges $99 a month for a hosted team of up to 12.
Last verified November 7, 2026, from each vendor's own pricing page. We are independent of every vendor here, with no affiliation or endorsement either way.
What do annotation vendors publish about price?
| Vendor | Pricing model | What is published | Minimum or trial |
|---|---|---|---|
| Kili Technologykili-technology.com, read Nov 7, 2026 | Subscription, custom | No-cost trial for 2 to 5 seats and at most 100 assets; Grow and Enterprise sold on custom terms. | Not published |
| Segments.aisegments.ai, read Nov 7, 2026 | Annual labeling hours (platform) | Core tier: 3,600+ labeling hours a year, which Segments.ai equates to $2.67 an hour. Fusion: 5,000+ hours. Enterprise: 150,000+ hours. 14-day trial. | Core tier needs 3,600+ hours a year |
| HumanSignal (Label Studio)humansignal.com, read Nov 7, 2026 | Per seat, monthly (platform) | Hosted Starter Cloud: $99/month with 12 users included, then $49/month for each added user. Enterprise priced on request. Community Edition: free, open source. | Not published |
| Label Your Datalabelyourdata.com, read Nov 7, 2026 | Per object, per entity or per annotator hour | $0.015+ per keypoint, $0.02+ per box, $0.02+ per NLP entity. Hourly work: usually $6 an annotator hour. | None stated; pilot is free |
| Labelboxdocs.labelbox.com, read Nov 7, 2026 | Usage units (LBUs) | Billed in LBUs, e.g. 1 per image row (Annotate) or 20 per row (live multimodal or LLM work). LBU dollar rate not shown. Free tier: 500 LBUs monthly. | Not published |
| SuperAnnotatesuperannotate.com, read Nov 7, 2026 | Plan tiers, custom | Three tiers (Starter, Pro, Enterprise) bundling 1K, 2.5K or 10K Orchestrate compute hours. Prices not shown. | Not published |
Listed vendors may reward us for introductions, which changes nothing printed here. Every figure is the vendor's own, dated, and none is a quote.
Quote only, with no figures on the pages we read: Dataloop, Sama, Scale AI, Encord, Appen, Centific, TELUS Digital (AI Data Solutions), Toloka, CloudFactory, Innodata, Snorkel AI, Keymakr.
Where these facts come from
- Kili Technology pricing: kili-technology.com, read Nov 7, 2026
- Segments.ai pricing: segments.ai, read Nov 7, 2026
- HumanSignal (Label Studio) pricing: humansignal.com, read Nov 7, 2026
- Label Your Data pricing: labelyourdata.com, read Nov 7, 2026
- Labelbox pricing: docs.labelbox.com, read Nov 7, 2026
- SuperAnnotate pricing: superannotate.com, read Nov 7, 2026
Which pricing models will you see?
- Per object or per task. A rate per box, keypoint, entity or item. Easy to budget when items are uniform; watch for dense scenes, where object counts swing.
- Per annotator hour. You pay for time. Fair for variable or exploratory work; ask for expected items per hour so you can convert it.
- Per seat. A platform license for each user. Your own team does the labeling, so labor is a separate cost.
- Usage units or hours. Platform use metered by data row or labeling hour, often with a yearly commitment.
- Custom quote or project price. A fixed price after scoping. Most managed services work this way.
The trade-offs between the first two are covered in per-hour vs per-task annotation pricing.
How do you turn an hourly rate into a cost per item?
Divide the hourly rate by the number of items an annotator finishes in an hour, then add the share of items that are reviewed or reworked. Cost per item = hourly rate ÷ items per hour × (1 + review and rework share). The calculator below does the arithmetic.
What makes annotation more expensive?
- Shape. A polygon or mask takes many times longer than a box.
- Density. Twenty objects per image cost about twenty times one, at per-object rates.
- Quality layers. Consensus (several annotators per item) or full review multiplies labor.
- Expertise. Domain specialists cost more per hour than general annotators.
- Speed. Short deadlines need bigger crews and more management.
- Security. Working inside your cloud, US-only staff or secure facilities add cost.
Put each of these in your brief so that quotes cover the same work. The RFP builder asks about all of them.
Questions about annotation pricing
How much does data annotation cost per image?
It depends on what is drawn on the image. At Label Your Data's published floor of two cents a box, an image holding 20 boxes starts near $0.40; keypoints start lower, at $0.015 each. Masks and dense scenes cost more. Treat any figure as a starting rate, not a quote.
What is a typical hourly rate for annotation?
Of the vendors we compared, only Label Your Data publishes an annotator rate: typically $6 per annotator hour for hourly projects. Expert annotators (engineers, lawyers, linguists) cost far more, and most vendors quote them on request.
Why do most annotation vendors not publish prices?
Because the cost depends on the task, the quality layers, the volume and the security setup. Vendors prefer to price after seeing a brief and a sample. A clear RFP gets you comparable quotes faster.
Is a platform cheaper than a managed service?
The platform fee usually is, but you supply and pay the annotators. Add your own team's hours, training and quality review before comparing it with a managed price that includes them.
Are the prices on this page quotes?
No. They are the vendors' own published figures on the date we read them. Your price depends on your data and the vendor's terms, and may be higher or lower.
Read next
Get a quote
A ZIP code, a reply address, then a few taps describing the work. We match the outline against what each vendor publishes and write back with names; no vendor hears about you until you pick one.
Keep data, samples, label guides and logins out of this form. The outline on its own is enough for us to work from.
How this site earns: as an introduction service, it may be paid by the annotation company you are introduced to.
Received. Next steps from our side:
- Your outline is set against each vendor's published modalities, minimums, security reports and pricing.
- We write to you naming the vendors we think fit, and why.
- Nothing goes to any vendor until you reply to say which introductions you want.