Facts checked on each company's own website on 24 October 2026.
The best data labeling company depends on the work. For ongoing human review, data operations, LLM evaluation and RLHF data, Deepen AI Services runs trained, in-house teams that keep the same people on your work every round, at a published $5 per person-hour for review and data work. For image, video and multi-sensor annotation, Deepen AI's enterprise programs, Scale AI, iMerit and Sama run scoped programs. For frontier-model data and expert networks, Scale AI, Surge AI, Mercor and Toloka supply AI labs and LLM teams. Appen cites coverage of more than 500 locales, and Labelbox and SuperAnnotate sell platforms for teams that label in-house.
The short answer: the right company for the work
| The work | Companies | What each offers |
|---|---|---|
| Human review and data operations inside your tools | Deepen AI Services, CloudFactory | Deepen AI Services publishes $5 per person-hour and sends a plan within 2 working days. CloudFactory prices on annual commitments. |
| LLM evaluation, RLHF data and expert work | Deepen AI Services, Scale AI, Surge AI, Mercor, Toloka | Deepen AI Services' trained raters work to your rubric, with the same people every round, and specialist work is quoted. Scale AI and Surge AI work with frontier labs, Toloka focuses on data for AI agents and LLMs, and Mercor runs an expert network. |
| Image, video and sensor annotation | Deepen AI, Scale AI, iMerit, Sama | Deepen AI labels across calibrated camera, LiDAR and radar. iMerit and Sama run managed annotation teams, and Scale AI runs large programs. |
| Labeling with a platform | Deepen AI Services in your tool, Labelbox, SuperAnnotate | Labelbox and SuperAnnotate sell the platform. Deepen AI Services' team can work inside the tool you already use. |
| Data in many languages | Deepen AI Services, Appen | Deepen AI Services quotes non-English work in your plan. Appen cites coverage of more than 500 locales. |
The companies compared
| Company | Known for | Pricing | How you start |
|---|---|---|---|
| Deepen AI Services | Trained in-house teams for review, data operations and AI evaluation, and Deepen AI's sensor data programs | $5 per person-hour for human review and data operations; other work quoted | A plan by email in 2 working days, then a two-week pilot |
| Scale AI | Data, evaluations and programs for AI labs, governments and the Fortune 500 | No prices on its site | Book an intro call |
| Appen | AI training data and evaluation in 500+ locales | No prices on its site | Talk to an expert |
| iMerit (part of EXL) | Image, video and LiDAR annotation, and generative AI services | No prices on its site | Talk to an expert or book a demo |
| Sama | Vision, multimodal and sensor annotation | Discussed in a first call | Talk to an expert |
| CloudFactory | Managed teams for AI data | Annual commitments, custom rates | A free sample analysis |
| Surge AI | Frontier data, RL environments and an expert workforce | No prices on its site | Contact its team |
| Mercor | Expert human data, evaluations and benchmarks | No client prices on its site | Book a demo or contact sales |
| Toloka | Training and evaluation data for AI agents and LLMs | No prices on its site | Talk to an expert |
| Label Your Data | Managed annotation with per-object prices | From $0.02 per bounding box; typically $6 an hour | A no-cost pilot |
| Labelbox | Training environments, robotics data and the Alignerr expert network | No prices on its homepage | Get started or talk to sales |
| SuperAnnotate | An AI data platform | Starter, Pro and Enterprise plans, priced by sales | Request a demo |
The companies in detail
1. Deepen AI Services: trained in-house teams, with published pricing
Deepen AI Services is Deepen AI's managed human data service. Trained, in-house teams work inside your tools, as your Ground Truth or A2I workforce, or on Deepen AI's own platform. Deepen AI has run trained data teams for automotive and robotics companies since 2017, and has more than 200 programs in production.
- Best for: Ongoing human review, data operations, LLM output review and RLHF data, plus image, video, 3D point-cloud and multi-sensor annotation through Deepen AI's enterprise programs.
- Who does the work: In-house teams trained and managed by Deepen AI, never crowdsourced, each with a named team lead.
- Pricing: $5 per person-hour for human review and data operations, with QA sampling and a weekly report included. AI evaluation, agent and robotics data are quoted in your plan, and enterprise programs are scoped in a statement of work.
- How you start: Describe the task and get a plan and a price by email within 2 working days, with no sales call. Then a two-week paid pilot on your real work.
- Sensor data: Labels made once across calibrated camera, LiDAR and radar, so each one projects into every sensor view. Exports to ASAM OpenLABEL, a standard Deepen AI co-authored, or your own schema. Sensor calibration as a managed service, a product or an API.
- Security: SOC 2 Type II, ISO 27001, ISO 9001 and TISAX, with work on accounts you control, under NDA.
- Track record: Deepen AI has delivered annotation and data services for Daimler Trucks, BMW, Bosch, Mercedes, Hexagon, Aptiv and Ford Otosan.
See the services, the enterprise programs and the pricing.
2. Scale AI: data and programs for AI labs and governments
Scale AI provides data, evaluations and AI programs to AI labs, governments and the Fortune 500, and runs large labeling programs on its own platform. Its site lists no prices, and you start by booking an intro call.
3. Appen: AI data across 500+ locales
Appen supplies AI training data, data annotation and model evaluation, and cites coverage of more than 500 locales, which makes it a common choice for multilingual work. Projects start with a conversation ("Talk to an expert"), and its site shows no prices.
4. iMerit: annotation and generative AI services, now part of EXL
iMerit annotates image, video and LiDAR data, including 3D sensor fusion, and offers generative AI services such as RLHF, evaluation and red teaming. It cites more than 25,000 domain experts in its Scholars program, and lists SOC 2, ISO 27001, GDPR and TISAX compliance. iMerit is now part of EXL. You start by talking to its team or booking a demo.
5. Sama: managed annotation by full-time teams
Sama offers expert-led labeling for vision, multimodal and sensor data, plus text annotation for generative AI, such as preference ranking. Sama says it creates full-time, fair-wage work, and it is a Certified B Corp. Pricing, scope and timelines are discussed in a first call.
6. CloudFactory: managed teams on annual commitments
CloudFactory offers managed teams for AI data and model monitoring. It starts with a free analysis of a sample of your data, then sends a proposal. There's no public rate card: pricing is built on annual spend commitments with custom rates.
7. Surge AI: data for frontier models
Surge AI focuses on data for frontier models: off-the-shelf datasets, post-training runs, RL environments and an expert workforce. Its site shows no prices, and you start through its contact page.
8. Mercor: experts for human data and evaluations
Mercor connects AI labs and enterprises with experts, such as physicians, lawyers and engineers, for human data, evaluations and benchmarks. It publishes what it pays experts by role, averaging $100 an hour, but not what clients pay. You start by booking a demo or contacting sales.
9. Toloka: data for AI agents and LLMs
Toloka focuses on training and evaluation data for AI agents and LLMs: agent demonstrations, preference data, red-teaming and expert work. Its site points businesses to a conversation with its experts and shows no prices.
10. Label Your Data: per-object pricing
Label Your Data is a managed annotation company that publishes its prices: from $0.02 per bounding box, $0.015 per keypoint and $0.02 per text entity, or typically $6 per annotator-hour for time-billed work. It offers a no-cost pilot.
11. Labelbox: environments, robotics data and experts
Labelbox now centers on training environments and evaluations for frontier AI, robotics data such as robot demonstrations, agents for enterprise work, and Alignerr, a network it describes as 2.6 million skill-tested experts in more than 200 domains. Its homepage shows no prices.
12. SuperAnnotate: an AI data platform
SuperAnnotate sells an AI data platform in Starter, Pro and Enterprise plans, with prices given through a demo or its sales team. Its pricing page doesn't list managed annotation services.
How to choose a data labeling company
- What's the work? Review and data operations, frontier-model data and sensor annotation are different markets. Start with companies that do your kind of work.
- Who does it? An in-house team, a crowd and an expert network give different consistency. Ask whether the same people stay on your work every week.
- Where does it happen? Some companies work inside your tools; others need your data on their platform.
- What will it really cost? Compare offers on the cost of each accepted label, including your own review time. See data labeling pricing in 2026.
- Does it pass your security review? Look for SOC 2 Type II and ISO 27001, TISAX for automotive data, an NDA before real data, and work on accounts you control.
- How do you start? A pilot on your real work shows more than any sales call.
Run a pilot before you commit
- Pick a sample of real items, including your hardest cases
- Add a small gold set with known answers
- Share your guidelines, with examples of work you'd accept and reject
- Agree what you'll measure: accepted items, questions raised, turnaround and cost per accepted item
- Ask for the same people in production as in the pilot
- Compare pilots on the cost of each accepted label, not the rate alone