How much does AI training data cost in 2026?
Last updated: 2026-11-09
Most AI training data providers price by custom quote. Three publish their pricing rules: Prolific (participant pay plus a platform fee), Toloka (pay only for work that passes QA) and Defined.ai (individual dataset licenses). The table below uses only what each provider's official site says. We did not request quotes.
Pricing model by provider
| Provider | Pricing model | What is published |
|---|---|---|
| Prolific | Pay as you go, or managed services | The participant reward you set plus a platform fee, usually 42.8% (corporate) or 33.3% (academic or nonprofit). Recommended pay is at least £9 / $12 per hour; the minimum is £6 / $8. No subscription. Managed services run under an SOW. |
| Toloka | Per-project price | Price is calculated after setup. No hidden fees, minimums or contracts. You pay only for work that passes QA. |
| Defined.ai | Dataset licenses plus quotes | Marketplace datasets are licensed individually. Custom work and volume discounts by quote. ACH, PO and SOW accepted. |
| Surge AI | Quote; off-the-shelf pay-if-it-helps | Says you can train on the full off-the-shelf dataset first and pay only if it improves your metrics. |
| Mercor | Quote | Shows a $90/hr average contracted expert rate. That is the rate experts are paid, not the price clients pay. |
| Scale AI | Quote | Demo or sales request only. |
| Turing | Quote | Dataset samples on request. |
| Labelbox | Quote | Sales contact; its former pricing page returns a 404. |
| Handshake AI | Quote | Scoping session. |
| Invisible Technologies | Quote | Demo request. |
| Snorkel AI | Quote | Dataset samples and sales contact. |
| TELUS Digital | Quote / off-the-shelf catalog | Contact sales. Engagements range from short MVPs to multi-year programs. Off-the-shelf datasets are available, but prices are not listed on the reviewed page. |
| Appen | Quote / catalog | Contact an expert; catalog datasets are available. |
| iMerit | Quote | Contact an expert or book a demo. |
| Sama | Quote | "Talk to an expert". GenAI work is backed by the SamaAssure quality guarantee. |
What drives the price
- Expertise: credentialed professionals cost far more per hour than generalist annotators.
- Task type: RL environments and long-horizon agent trajectories cost more than simple labels.
- Quality assurance: multi-stage review, adjudication and calibration add cost but reduce rework.
- Volume and turnaround: rush deadlines raise prices; volume can bring discounts (Defined.ai says it offers them by quote).
- Security: SOC 2, ISO certifications, NDAs and regional data storage requirements can affect the price.
How to compare quotes
Send every vendor the same specification: modality, domain expertise, number of items, QA method, acceptance criteria and deadline. Then run a small paid pilot and compare how much your own evals improve per dollar spent.