Get started with 1,000 free operations per month.
Start building right away.
For businesses ramping with AI.
Most advanced, higher-volume.
1,000 free operations per month
Taxes may apply
MLOps without DevOps. Accelerate your developers, data scientists and business operators.
For the most demanding enterprise use cases.
Infrastructure designed for demanding workflows.
Use your AWS, Google or Azure cloud. Even bare metal and air gapped military-grade appliances.
Dedicated data, training and inference clusters available.
Role base access levels and single sign on for large teams.
Highest discounts as usage grows, so benefits increase with your success.
Compare Plans |
Community |
Essential |
Professional |
Enterprise |
|
Minimum monthly usage Download price sheet Each paid plan allows users to consume operations at no additional cost up to the package price. Additional operations are charged at the respective unit price per operation. See FAQ for more information (*).
|
1k operations 1k input objects |
$30 |
$300 |
|
Data Store |
||||
|
Applications An application is like a project.
|
1 |
2 |
5 |
Unlimited |
|
Image inputs We support JPEG, PNG, TIFF, BMP, WEBP. Enterprise plans support additional custom formats.
|
||||
|
Videos inputs We support AVI, MP4, WMV, MOV, GIF, 3GPP. Enterprise plans support additional custom formats.
|
||||
|
Text inputs We support ASCII text.
|
||||
|
Data collectors Connect data collectors to your workflows to save your production inputs back into your project for incremental training or auto-labeling.
|
||||
|
App duplication Duplicate your entire app, including its inputs, concepts and models.
|
||||
|
App reindexing Change the default workflow for applications to any valid workflow and optionally reindex existing inputs.
|
||||
Scribe Label |
||||
|
Concepts per app Add unlimited concepts to your apps.
|
Unlimited |
Unlimited |
Unlimited |
Unlimited |
|
Labeled concepts Add unlimited annotations to your apps.
|
Unlimited |
Unlimited |
Unlimited |
Unlimited |
|
Single label classification Categorize images, video and text into one category out of many.
|
||||
|
Multi label classification Categorize images, video and text into multiple categories.
|
||||
|
Bounding boxes Draw and annotate bounding boxes around objects of interest.
|
||||
|
Multi-point polygons Plot points on each vertex of a target object.
|
||||
|
Video interpolation Annotate objects across video frames using intelligent frame interpolation.
|
||||
|
AI assisted labeling AI-assisted labeling assigns concepts using pre-built or custom model workflows for faster dataset annotation at scale.
|
||||
|
Labeling task management A variety of workforce management tools that help your company to distribute and manage labeling tasks.
|
||||
|
Task assignment rules Support for full and partitioned worker strategies. In a full worker strategy, each worker labels all inputs selected in the input source. In a partitioned worker strategy, all tasks are divided evenly across workers.
|
||||
|
Optimized task review Powerful Uis optimized for large jobs.
|
||||
|
Consensus review Automatically approve labels when enough worker annotations match.
|
||||
Spacetime Search |
||||
|
Search by concept Retrieve images from your dataset by searching for the concepts identified by AI models.Ê
|
||||
|
Search by image Retrieve visually similar images or videos for a given input.
|
||||
|
Search by metadata Retrieve images based on custom metadata.
|
||||
|
Search by geo-location Retrieve images based on longitude, latitude and radius.
|
||||
Mesh Workflow |
||||
|
Trained models per app The number of trained models per app.
|
1 |
5 |
20 |
|
|
Custom workflows The number of custom workflows per app.
|
1 |
3 |
10 |
Unlimited |
|
Models in workflow graph The number of model nodes within a workflow graph.
|
3 |
6 |
9 |
Unlimited |
|
Image classification models Create context-based visual concept classification models.
|
||||
|
Image detection models Create context-based visual object detection models.
|
||||
|
Text classification models Create context-based text / document classification models.
|
||||
|
Image cropper Crop the input image according to each input region that is present in the input.
|
||||
|
Concept thresholder Use a threshold and an operator (>, >=, =, <=, or <) to route the outputs of classification models in your workflows.
|
||||
|
Random sampler Randomly sample inputs allowed to pass through as outputs.
|
||||
|
Annotation writer Write the input data to the database in the form of an annotation with a specified status as if a specific user created the annotation.
|
||||
Enlight Train |
||||
|
Model versions per model |
2 |
5 |
40 |
|
|
Custom model training Context-based model training leveraging a pre-built deep trained model. You can also context train using your own deep trained model if you are on a plan that includes deep training.
|
||||
|
Deep training Train every layer in a convolutional network from the ground up.
|
||||
|
Model evaluation |
||||
|
Evaluation summaries Evaluation overall model performance. Identify high-performing and low-performing concepts.
|
||||
|
K-split confusion matrices |
||||
|
ROC AUC ConceptÕs prediction performance score, defined the area under the receiver operating characteristic curve.
|
||||
Armada Predict |
||||
|
Clarifai pre-built model predictions Dozens of high performance pre-built models.
|
||||
|
Custom model serving Predictions from your own custom models.
|
||||
|
Shared inference servers Dynamically-scaling inference clusters.
|
||||
|
Dedicated inference servers Dedicated nodes for predictions.
|
||||
Collaborate |
||||
|
Collaborators The number of distinct collaborators.
|
2 |
5 |
||
|
Fine grained API keys Fine-grained control over the data exposed through your app.
|
||||
|
Additional collaborators Additional collaborators across apps can be purchased as an add-on by contacting sales.
|
||||
|
Single Sign On Use a custom identity provider to connect to Clarifai.
|
||||
Support |
||||
|
Portal and API documentation docs.clarifai.com documents APIs, user interfaces and best practices.
|
||||
|
Community Slack and Stackoverflow Access our communities via Slack or Stackoverflow.
|
||||
|
Priority email support Priority email queues.
|
||||
|
Service level agreement Service level agreements for enterprise grade use cases.
|
||||
|
Early access program Get early access to our latest features.
|
||||
|
Expert education and onboarding Custom training sessions by Clarifai ML experts.
|
||||
|
Dedicated solution architect Consult with Clarifai ML experts on your use case.
|
See why the IDC MarketScape named Clarifai a Leader in Computer Vision AI Software Platforms.
Outsource your data labeling tasks to a team of specialized data annotators and researchers. We combine cutting edge AI automation on a secure AI platform to deliver you the highest quality training datasets you need for the most complex use cases. Contact us to learn about additional volume discounts.
Learn more our data labeling services
| First 500,000 annotations | 500,000+ annotations | |
| Visual Classification | $0.05 per image or frame and $0.05 per annotation | Volume discounts |
| Bounding Boxes | $0.05 per image or frame and $0.05 per annotation | Volume discounts |
| Text | $0.05 per document and $0.05 per annotation | Volume discounts |
Premium subscription packages have a minimum usage commitment. Each operation has a price. Each indexed input object has a price. Depending on the plan, you will not be billed more than the minimum monthly price until your usage surpasses it.
The Essential plan lets users do a max of 300k operations each month. The Professional plan lets users do a max of 1M operations each month. Similarly, the Essential plan lets users store a max of 300k inputs, and the Professional plan lets users store a max of 1M inputs. Since there are different prices for ops and inputs, there are additional charges for any inputs or operations beyond the package price.
You can cancel your subscription any time by downgrading to the community plan. Once you downgrade, you are allowed inputs and operations will be bound by our community plan. Contact support@clarifai.com.
Clarifai's Enlight custom model training lets you take advantage of our pre-trained models, which are trained on millions of inputs. This lets you train high-performing models that leverage the understanding from our pre-trained models using a very small number of inputs. On the other hand, with deep training you can build an entirely custom neural network for your application from the ground-up, and thus requires a higher number of inputs. Deep training is extremely powerful, particularly when you have unique data and you need to maximize model accuracy. Learn more about deep training.
Custom plans are available and scale as your business grows.
Download our price sheet
Clarifai’s team of experts are here to help you every step of the way as you integrate computer vision AI into your technology stack. We offer Scribe LabelForce Data Labeling service as well as Enlight ModelForce for custom modeling building.
Please contact us below for a detailed Statement of Work and level of effort.
Learn more about Clarifai
Schedule a demo
Discuss our solution options
1,000 free operations per month.
If you are encountering a technical or payment issue, the customer support team will be happy to assist you.
Contact support
