Data Annotation Tools

Arrow keys to move, Enter to open, Esc to close

Data Labeling Tools and Platforms

Data labeling tools are the software a team uses to label training data together, with one place to hand out work, review it and export it. This page lists the 19 tools whose docs describe a real review step, from free self-hosted projects to platforms that only quote through sales.

Facts checked Nov 8, 2026

The Hubble eXtreme Deep Field with 250 galaxies, each inside a detection box
Labeling at scale: 250 galaxies boxed by code in a NASA public-domain image, the kind of first pass a person then reviews.

19 data labeling platforms with a review step

Facts checked against each tool's own docs, pricing page or repository on Nov 8, 2026. Order: most of our 8 documented checks first. No tool paid for its place. How we compare.

  1. 1

    Label Studio

    HumanSignal · Open core (Apache-2.0)

    8/8 checks

    Review
    Review stage, Consensus, Gold or honeypot items, Agreement scores
    Runs
    On your own server, on private infrastructure (VPC or on-premises), or in the vendor's cloudLast release Sep 29, 2026
    Price
    Community Edition: free and open source (Apache-2.0), self-hosted. Starter Cloud: $99 a month plus $49 a month per extra user, up to 12 users. Enterprise: by quote.
    The catch
    Reviewer assignment, agreement metrics and roles are not in the free Community Edition; they start with Starter Cloud.
  2. 2

    Labellerr

    Tensor Matics, Inc. · Commercial

    8/8 checks

    Review
    Review stage, Agreement scores, Roles
    Runs
    On private infrastructure (VPC or on-premises) or in the vendor's cloud
    Price
    Free Researcher plan for students and researchers (by application): 2,500 data credits, 1 seat. Pro: $9,999 a year. Enterprise: by quote.
    The catch
    DICOM, LiDAR and NIfTI files are listed only on the custom-priced Enterprise plan, and its FAQ quotes a Pro price of $499 a month while the pricing table says $9,999 a year.
  3. 3

    Datasaur Data Studio

    Datasaur · Commercial

    8/8 checks

    Review
    Review stage, Consensus, Agreement scores, Roles
    Runs
    On your own server or in the vendor's cloud
    Price
    Free: personal workspace, 5,000 labels a year. Starter: from $5,000 a year through sales, up to 3 users. Growth: from $24,000 a year.
    The catch
    Paid plans go through sales, self-hosting is Enterprise only, and Datasaur's main site now leads with private LLM projects.
  4. 4

    CVAT

    CVAT.ai Corporation · Open core (MIT)

    8/8 checks

    Review
    Review stage, Consensus, Gold or honeypot items
    Runs
    On your own server, on private infrastructure (VPC or on-premises), or in the vendor's cloudLast release Nov 5, 2026
    Price
    Community Edition: free and open source (MIT), self-hosted. CVAT Online: free plan with 1 GB and 3 tasks; Solo $33 a month, or $23 a month billed yearly; Team $33 per user a month, 2 seats minimum.
    The catch
    CVAT's own FAQ says the free self-hosted edition has no automated QA and no SSO, and that CVAT holds no SOC 2 or ISO 27001 certification.
  5. 5

    UBIAI

    UBIAI · Commercial

    8/8 checks

    Review
    Review stage, Agreement scores
    Runs
    On your own server, on private infrastructure (VPC or on-premises), or in the vendor's cloud
    Price
    Start: $0 a month, one user, 100 documents a month. Essential: $74 a month. Growth: $299 a month.
    The catch
    LLM pre-labeling needs your own OpenAI API key, and its annotation docs show they were last updated about two years ago.
  6. 6

    Kili Technology

    Kili Technology · Commercial

    7/8 checks

    Review
    Review stage, Consensus, Gold or honeypot items, Roles
    Runs
    On private infrastructure (VPC or on-premises) or in the vendor's cloud
    Price
    Free Trial plan: $0 a month, 2 to 5 seats, 100 assets (5 for video). Grow and Enterprise: custom pricing, no numbers printed.
    The catch
    Each project tops out at 25,000 assets, so bigger datasets get split across projects.
  7. 7

    BasicAI

    BasicAI · Commercial

    7/8 checks

    Review
    Review stage, Roles
    Runs
    On private infrastructure (VPC or on-premises)
    Price
    Private-cloud deployment from $6,600 a year, sized per company. No free plan on the pricing page.
    The catch
    Self-serve cloud plans are gone from the pricing page, while its docs still describe a free cloud plan, so confirm what you're buying.
  8. 8

    Supervisely

    Supervisely · Commercial

    7/8 checks

    Review
    Review stage, Consensus
    Runs
    On your own server, on private infrastructure (VPC or on-premises), or in the vendor's cloud
    Price
    Community: free, 2 members, 5 GB, hosted in Germany. Pro: from €199 a month with 3 members, €85 per extra member. Enterprise: by quote.
    The catch
    Pro's base price leaves out the advanced toolboxes: point cloud, video and medical tools at full limits are paid add-ons, and point clouds alone add €399.
  9. 9

    INCEpTION

    Eclipse Foundation incubating project, originally developed at the UKP Lab, TU Darmstadt · Open source (Apache-2.0)

    7/8 checks

    Now an Eclipse Foundation project, named Eclipse INCEpTION.

    Review
    Review stage, Consensus, Agreement scores
    Runs
    On your computer, on your own server, or on private infrastructure (VPC or on-premises)Last release Oct 27, 2026
    Price
    Free and open source (Apache-2.0), self-hosted or as a desktop app.
    The catch
    The PDF editor needs a text layer, so scanned PDFs must be OCRed first, and their text can sit slightly off the page image.
  10. 10

    Roboflow

    Roboflow · Cloud service

    7/8 checks

    Review
    Review stage, Roles
    Runs
    In the vendor's cloud
    Price
    Free Tier: $0, 10 credits a month. Core: $39 a month for a credit pack. Credits pay for storage, AI labeling, training and deployment.
    The catch
    New workspaces have data licensing and model training on private projects switched on by default; an admin has to opt out. Review Mode and roles need the Teams add-on.
  11. 11

    Encord

    Encord (Cord Technologies, Inc.) · Commercial

    6/8 checks

    Review
    Review stage, Consensus, Gold or honeypot items
    Runs
    On private infrastructure (VPC or on-premises) or in the vendor's cloud
    Price
    No prices printed for the Starter, Team or Enterprise plans. No free plan is listed.
    The catch
    Point clouds, DICOM and NIfTI are add-ons on the Team plan and come with Enterprise, where VPC and on-prem installs are add-ons too.
  12. 12

    Dataloop

    Dataloop Ltd., part of Dell Technologies · Commercial

    6/8 checks

    Renamed Dell Data Orchestration Engine on March 30, 2026.

    Review
    Review stage, Consensus, Gold or honeypot items
    Runs
    On private infrastructure (VPC or on-premises) or in the vendor's cloud
    Price
    No public prices: billing is metered on time spent in the labeling tool, stored items and compute. The only free option is an academic license.
    The catch
    Its COCO, YOLO and Pascal VOC exports cover images only; everything else exports as its own JSON.
  13. 13

    Prodigy

    Explosion · Paid license

    6/8 checks

    Review
    Review stage, Consensus, Agreement scores
    Runs
    On your computer, on your own server, or on private infrastructure (VPC or on-premises)
    Price
    Personal license: $390 once. Company: $490 per seat, sold in packs of 5. Both include 12 months of upgrades. No free version.
    The catch
    No user accounts or project management; its docs say it leaves that kind of feature out on purpose.
  14. 14

    Segments.ai

    Segments.ai (site logo reads 'Segments.ai by Uber') · Cloud service

    6/8 checks

    Its site now brands it Segments.ai by Uber.

    Review
    Review stage
    Runs
    In the vendor's cloud
    Price
    No free plan, a 14-day trial on Core. Core: custom quote from 3,600 labeling hours a year (equal to $2.67 an hour), unlimited seats.
    The catch
    The entry Core plan caps each point cloud at 500,000 points; bigger scans need the Fusion plan.
  15. 15

    Audino

    MIDAS, IIIT Delhi · Source available (CC)

    6/8 checks

    Review
    Gold or honeypot items
    Runs
    On your own serverLast release Feb 3, 2026
    Price
    Free. Version 2.0 is licensed for non-commercial use only (CC BY-NC 4.0); the older v1 branch is MIT.
    The catch
    Since February 2026 the current version (v2.0) is licensed for non-commercial use only; the older v1 branch is still MIT.
  16. 16

    SuperAnnotate

    SuperAnnotate · Commercial

    5/8 checks

    Review
    Review stage, Roles
    Runs
    In the vendor's cloud
    Price
    No prices printed: Starter, Pro and Enterprise all go through a demo request. No free plan is listed.
    The catch
    Cloud only, and labels export as SuperAnnotate JSON, JSONL or CSV; COCO needs its Python SDK converter.
  17. 17

    Labelbox

    Labelbox, Inc. · Commercial

    5/8 checks

    Review
    Review stage, Consensus, Gold or honeypot items
    Runs
    In the vendor's cloud
    Price
    Free plan: 500 Labelbox Units (LBU) a month, up to 30 users. Starter: $0.10 per LBU, and a labeled image costs 1 LBU. Enterprise: by quote. Prices come from its docs; labelbox.com/pricing returned a 404 on Nov 8, 2026.
    The catch
    Cloud only, with data stored in the US; Labelbox retired its on-premise version, moving the last customers off by August 2023, and labels export as Labelbox JSON.
  18. 18

    V7 Darwin

    V7 · Commercial

    5/8 checks

    Review
    Review stage, Consensus, Agreement scores, Roles
    Runs
    In the vendor's cloud
    Price
    Custom yearly quote: a platform fee plus user licenses plus data processing. The Darwin site mentions a free tier but prints no limits.
    The catch
    Its COCO, YOLO, Pascal VOC and CVAT XML exports don't carry video labels; only Darwin JSON covers every annotation type.
  19. 19

    VisionRepo

    Averroes · Cloud service

    5/8 checks

    Review
    Review stage, Consensus, Roles
    Runs
    In the vendor's cloud
    Price
    Free: $0, one seat, one project, 2 GB. Pro: $20 a month for 3 seats. Ultra: $200 a month for 7 seats.
    The catch
    Labeling itself is metered: every label you create or remove counts against a daily quota that depends on the plan.

Data labeling platform or data labeling tool?

A data labeling platform and a data labeling tool overlap, but platform usually means more than the labeling screen: user roles, review queues, quality scores, storage connections and an API, often with model training or an outside workforce sold on top. A tool can be a one-person app such as Labelme or X-AnyLabeling, which suit a single labeler and have no reviewer roles; the full list of data annotation tools includes those too. If more than one person will label, you want platform features, so the list above only includes tools whose docs describe a review step.

Data labeling software pricing

Data labeling software pricing comes in four shapes, and the same dataset can cost very different amounts under each. The examples below are copied from each vendor's pricing page or docs on November 8, 2026.

Pricing modelHow it billsExamples, as printed
Per seatA monthly or yearly price per userLabel Studio Starter Cloud: $99 a month plus $49 per extra user. CVAT Team: $33 per user a month. Supervisely Pro: from €199 a month with 3 members.
Usage unitsCredits or units per labeled item, stored row or computeLabelbox Starter: $0.10 per Labelbox Unit. Roboflow Core: $39 a month for a credit pack.
Labeling hoursTime spent in the labeling toolSegments.ai Core: from 3,600 labeling hours a year, equal to $2.67 an hour. Dataloop meters time in its labeling tool.
Yearly contract or quoteA figure from salesLabellerr Pro: $9,999 a year. Datasaur Starter: from $5,000 a year. SuperAnnotate, Encord, V7 Darwin and Kili Technology print no plan prices.

5 of the tools on this site print no price for any paid plan, so plan on a sales call before you can compare them. With usage pricing, ask what a full project costs, since video, PDF pages and stored rows can bill differently from a single labeled image.

Automated data labeling

Automated data labeling lets a model label the easy items so people only check and fix the rest. Platforms do it in three ways: they run or import a model's predictions as pre-labels, they let a promptable model such as SAM turn a click into an outline, or they use active learning to send the items the model is least sure about to people first.

  • CVAT's AI tools prompt SAM 2 and SAM 3 for masks, auto-label with Hugging Face or Roboflow models, and track objects through video, though its own FAQ says the free self-hosted edition has no SAM integration.
  • Labellerr runs Auto Label Jobs that combine active learning with zero-shot models.
  • VisionRepo auto-labels with a custom model trained on your own labels on Pro and above, and Averroes says about 40 labeled examples are enough to start.
  • Amazon SageMaker Ground Truth offered automated data labeling on AWS, but its docs now say it is closed to new customers.

Free data labeling software

Free data labeling software splits into open-source projects you host yourself and free plans on hosted platforms. The open-source side includes CVAT's Community edition with manual review and INCEpTION with curation and agreement scores, plus doccano and Argilla, which let several people label one project but document no separate review stage. The hosted side includes CVAT Online's free plan (1 GB, 3 tasks), Supervisely Community (2 members, 5 GB), Labelbox's 500 free units a month, VisionRepo Free (one seat), Kili Technology's Free Trial plan (100 assets), Datasaur's free workspace (5,000 labels a year) and UBIAI Start (100 documents a month).

How to choose a data labeling platform for a team

Choosing a data labeling platform for a team comes down to five questions, and the answers rule tools out fast.

  1. Where must the data live? Labelbox stores data in the US and V7 Darwin in the EU; Roboflow, SuperAnnotate and Segments.ai list no self-hosted edition, though some platforms can read files straight from your own cloud storage. CVAT and Label Studio can run on your server, Supervisely, Kili Technology and Encord offer on-premises installs on their top plans, and Dataloop added one in July 2026.
  2. How will you check quality? Look for review stages, consensus between labelers and gold-standard or honeypot items, and check which plan includes them: Label Studio keeps reviewer assignment out of its free edition, and Roboflow needs a Teams add-on for Review Mode.
  3. What does it export? Make sure your training code can read the output without a converter.
  4. How does the bill grow? Per-seat plans grow with your team, usage plans with your data and hours plans with labeling time.
  5. What happens to your data? Roboflow switches data licensing and model training on by default for private projects in new workspaces, while Supervisely's security page says customer data is not used to train models.

Data labeling tools FAQ

These data labeling tools FAQ answers come from the vendor facts behind the list above.

What is a data labeling tool?

A data labeling tool is software for adding the labels a machine learning model learns from, such as boxes on images, categories on documents or entity tags on text. Team platforms add roles, review steps and exports on top of the labeling screen.

What is the best data labeling platform?

The best data labeling platform is the one that handles your data where it has to live. On our documented checks, Label Studio, Labellerr, Datasaur Data Studio, CVAT and UBIAI each pass all 8; for a managed cloud with review workflows, also compare Labelbox, Encord, V7 Darwin and SuperAnnotate, which print fewer facts about price.

Is there free data labeling software?

Yes, free data labeling software includes open-source projects such as CVAT, Label Studio's Community Edition, doccano and INCEpTION, and free plans on hosted platforms such as CVAT Online, Supervisely Community and Labelbox.

How much do data labeling tools cost?

Data labeling tools cost from nothing for open-source software you host, to about $20 to $100 a month for small hosted plans such as VisionRepo Pro or Label Studio Starter Cloud, to yearly contracts such as Labellerr Pro at $9,999. Several enterprise platforms print no price at all.

Can data labeling be automated?

Data labeling can be partly automated: a model drafts labels and people review them. Fully automatic labels still need spot checks, because a model's mistakes repeat across the whole dataset.

Sources for this comparison

The sources for this comparison are the tools' own pages: 307 pages from 19 tools, read on Nov 8, 2026. They are listed as plain text, because we link a tool only when it gives us a referral link we disclose.

Show the 307 source pages
  • Label Studio: labelstud.io, api.github.com/repos/HumanSignal/label-studio, api.github.com/repos/HumanSignal/label-studio/releases/latest, labelstud.io/guide/label_studio_compare, raw.githubusercontent.com/HumanSignal/label-studio/develop/README.md, humansignal.com/pricing and 4 more
  • Labellerr: labellerr.com/pricing, docs.labellerr.com/changelog.md, docs.labellerr.com/product-releases/september-2026.md, labellerr.com/faq, labellerr.com/enterprise, docs.labellerr.com/getting-started/introduction.md and 24 more
  • Datasaur Data Studio: datasaur.ai, datasaur.ai/pricing, datasaur.ai/studio, docs.datasaur.ai, datasaur.ai/studio/pricing, docs.datasaur.ai/deployment/self-hosted.md and 8 more
  • CVAT: api.github.com/repos/cvat-ai/cvat/releases/latest, cvat.ai/resources/changelog/2026-09-24, docs.cvat.ai/docs/annotation/audio-editor, docs.cvat.ai/docs/annotation/manual-annotation/modes/3d-object-annotation, raw.githubusercontent.com/cvat-ai/cvat/develop/README.md, cvat.ai/pricing/cvat-online and 9 more
  • UBIAI: ubiai.tools, ubiai.gitbook.io/ubiai-documentation, ubiai.tools/pricing-ubiaiv2, ubiai.gitbook.io/ubiai-documentation/upload-documents.md, ubiai.tools/r1-ocr-annotation-feature, ubiai.tools/r1-auto-labeling and 5 more
  • Kili Technology: kili-technology.com/pricing, docs.kili-technology.com/changelog/november-4th-2026.md, kili-technology.com, docs.kili-technology.com/docs/labeling-audio-assets.md, docs.kili-technology.com/docs/labeling-llm-data.md, docs.kili-technology.com/docs/hosting.md and 29 more
  • BasicAI: basic.ai/basicai-data-annotation-platform/pricing, docs.basic.ai/docs/subscription.md, docs.basic.ai/changelog, basic.ai, docs.basic.ai/docs/gen-ai.md, docs.basic.ai/docs/class.md and 9 more
  • Supervisely: supervisely.com/pricing, supervisely.com/security, supervisely.com/platform, supervisely.com/ai-assisted-labeling, supervisely.com, supervisely.com/labeling-workflow and 1 more
  • INCEpTION: api.github.com/repos/inception-project/inception/releases/latest, eclipse-inception.github.io, inception-project.github.io, api.github.com/repos/inception-project/inception, eclipse-inception.github.io/eclipse, eclipse-inception.github.io/downloads and 1 more
  • Roboflow: docs.roboflow.com/platform/billing-and-plans/plans.md, roboflow.com/annotate, docs.roboflow.com/datasets/annotate/annotate/annotation-tools.md, docs.roboflow.com/datasets/annotate/annotate/annotate-video-segments.md, docs.roboflow.com/datasets/annotate/annotate/ai-labeling.md, docs.roboflow.com/datasets/annotate/annotate/ai-labeling/smart-select.md and 8 more
  • Encord: encord.com/pricing, docs.encord.com/release-notes/releasenotes-2026.md, docs.encord.com/platform-documentation/Other/deployment-options.md, docs.encord.com/platform-documentation/General/general-supported-data.md, docs.encord.com/platform-documentation/Annotate/annotate-ontologies/annotate-ontologies.md, docs.encord.com/platform-documentation/Annotate/automated-labeling/annotate-sam.md and 11 more
  • Dataloop: docs.dataloop.ai/docs/platform-release-notes.md, dataloop.ai, dataloop.ai/platform, docs.dataloop.ai/docs/home.md, dataloop.ai/solutions/data-annotation, docs.dataloop.ai/docs/data-annotation-overview.md and 24 more
  • Prodigy: prodi.gy/docs/changelog, prodi.gy/docs, prodi.gy/buy, prodi.gy/docs/install, prodi.gy/docs/deployment, prodi.gy/docs/recipes and 3 more
  • Segments.ai: segments.ai/pricing, docs.segments.ai/reference/sample-types.md, segments.ai/data-labeling/multi-sensor, segments.ai/data-labeling/3d-point-cloud, docs.segments.ai/reference/label-types.md, docs.segments.ai and 6 more
  • Audino: raw.githubusercontent.com/midas-research/audino/main/README.md, api.github.com/repos/midas-research/audino, api.github.com/repos/midas-research/audino/commits?per_page=10, api.github.com/repos/midas-research/audino/releases/latest, raw.githubusercontent.com/midas-research/audino/main/LICENSE, raw.githubusercontent.com/midas-research/audino/master/LICENSE and 4 more
  • SuperAnnotate: doc.superannotate.com/changelog/october-11-2026, doc.superannotate.com/docs/faq-data, trust.superannotate.com, superannotate.com/blog/superannotate-nvidia-enterprise-ai-reference-architecture, superannotate.com/pricing, superannotateai.github.io/custom-tools-docs and 15 more
  • Labelbox: docs.labelbox.com/horizon/release-notes.md, labelbox.com, docs.labelbox.com/llms.txt, labelbox.com/products/horizon, docs.labelbox.com/horizon/guides/deprecations.md, docs.labelbox.com/horizon/guides/image-editor.md and 18 more
  • V7 Darwin: v7darwin.com/changelog/audio-annotations, trust.v7labs.com, v7darwin.com/changelog, v7darwin.com/pricing, v7labs.com/pricing, v7darwin.com/security and 17 more
  • VisionRepo: averroes.ai/visionrepo/pricing, averroes.ai/visionrepo, averroes.ai/deployment