Encord
encord.com
- Web
- API
- Self-hosted
What Encord is
Encord is a multimodal data infrastructure layer for AI teams, covering work from pre-training and post-training through deployment. It provides human-in-the-loop labeling and review for image, video, audio, text, DICOM, LiDAR, geospatial, ECG, and other data. Image tools include boxes, polylines, polygons, primitives, keypoints, bitmasks, and native SAM 2 support, with model-assisted labeling. Dataset curation uses metadata and attribute filters, embeddings-based search, and natural-language search. Encord also supports model evaluation, rubric-based evaluation, RLHF, and pairwise model comparison. Projects, datasets, and annotations are available through an API and SDK, with AWS S3, Azure Data Lake Storage, and Google Cloud Storage integrations named. Deployment options include US or EU regions, private VPC, and on-premises. Pricing is on request, with no free plan. It is available through API, self-hosted, and web deployments.
Who it suits
Encord suits AI teams preparing multimodal datasets, evaluating models, and connecting annotation work through APIs. Its listed plans address individuals, small teams, scaling teams, and enterprises.
Good to know, and what to watch
- Supports many multimodal data types
- Includes model-assisted labeling
- Offers dataset search and curation tools
- Provides API and SDK access
- Supports private VPC and on-premises deployment
- Pricing is available only on request
- There is no free plan
- Some enterprise capabilities require Enterprise
Plans
Every plan and priceEnterprise
Starter
Team
Pricing on request
The verdict
Encord brings labeling, curation, evaluation, and deployment-oriented data work into one platform. It is most suitable when multimodal annotation and enterprise deployment options are priorities.
What the maker publishes
- Annotation
- The platform supports human-in-the-loop labeling and review for image, video, audio, text, DICOM, LiDAR, geospatial, ECG, and other data. source
- Annotation tools
- Image annotation includes boxes, polylines, polygons, primitives, keypoints, bitmasks, and native SAM 2 support. source
- Company
- The company page says Encord powers multimodal AI development from training to real-world deployment and reports $110M in total funding and 300+ teams. source
- Compliance
- Encord states it has completed a SOC 2 examination and describes itself as HIPAA compliant and GDPR compliant. source
- Curation
- Encord offers metadata and attribute filtering, embeddings-based search, and natural-language search for dataset curation. source
- Data handling
- With cloud integration, Encord says its platform does not store customer data by design; direct platform storage is also available, using Google Storage. source
- Data residency
- Encord offers US and EU deployment options as well as private VPC and on-premises deployments. source
- Evaluation
- The platform supports model evaluation, rubric-based evaluation, RLHF, and pairwise model comparison. source
- Integrations
- Cloud storage integrations named on the site include AWS S3, Azure Data Lake Storage, and Google Cloud Storage; users can access projects, datasets, and annotations through API/SDK. source
- Intended users
- The pricing page positions Starter for individuals and small teams prototyping AI applications, Team for teams scaling a few applications, and Enterprise for companies shipping multiple applications across teams. source
- Product
- Encord describes itself as a multimodal data infrastructure layer for AI teams, spanning pre- and post-training through deployment. source
- Security
- The security page says data at rest is encrypted with AES-256 and data in transit uses HTTPS over TLS, including TLS 1.3 support. source
- Support
- The listed plans include self-serve support for Starter, onboarding support for Team, and enterprise SLA and support for Enterprise. source
Checked against the maker's pages on 8 Oct 2026.