Leverage our next-generation studio to train, tune, validate and deploy AI models, and fit-for-purpose data store for scaling AI workloads based on an open data lakehouse architecture.
Get started with your governance strategy to direct, manage and monitor your organization’s AI activities.
Check out the different pricing tiers for the watsonx subscription to get started building AI applications for production‐grade use, built on top of our hybrid cloud platform.
watsonx.ai
Model inference based on Resource Unit (per 1000 tokens)2
ML functionality based on Capacity Unit Hours (per tool or runtime)1
ML functionality (20 CUH limit /month)1
Inferencing (50000 RU/month)2
Prompt Lab
Open source models
IBM-developed watsonx models
Synthetic Data Generator
ML functionality
Inferencing
Prompt Lab
Open source models
IBM-developed watsonx models
Synthetic Data Generator
Prompt tuning
USD 1050 per month - Tier fee
ML functionality
Inferencing
Prompt Lab
Open source models
IBM-developed watsonx models
Synthetic Data Generator
Prompt tuning
Custom foundation models
Model hosting
watsonx.data
Supporting services - USD 3 per hour
IBM Cloud
Presto:
Cache optimized node - USD 2.80 per hour.
Compute optimized node - USD 6.50 per hour.
- Small: 4 cache nodes
- Medium: 7 cache nodes
- Large: 12 cache nodes
Presto C++:
Cache optimized node - USD 4.20 per hour.
Compute optimized node - USD 9.75 per hour.
- Small: 4 cache nodes
- Medium: 7 cache nodes
- Large: 12 cache nodes
Spark:
- Small: 8vCPU, 32Gb Memory, and 300Gb secondary storage. USD 1.00 per hour.
- Medium: 16vCPU, 64Gb Memory, and 900GB secondary storage. USD 2.00 per hour.
- Large: 32vCPU, 128Gb Memory and 1600GB secondary storage. USD 3.69 per hour.
Milvus:
- Small: 10 Million vectors, Index Type- IVF_FLAT, Index Parameters – 64, Segment size (1024), Dimensions (at 384). USD 2.75 per hour.
- Medium: 50 Million vectors, Index Type- IVF_FLAT, Index Parameters – 64, Segment size (1024), Dimensions (at 384) USD 9.75 per hour.
- Large: 100 Million vectors, Index Type- IVF_FLAT, Index Parameters – 64, Segment size (1024), Dimensions (at 384). USD 16.50 per hour.
Hive metastore and Iceberg catalog
Infrastructure manager and query
Presto, Presto C++, Spark, Milvus
Db2 Warehouse and Netezza integrations
Hive metastore and Iceberg catalog
Infrastructure manager and query
Presto, Presto C++, Spark, Milvus
Db2 Warehouse and Netezza integrations
watsonx.governance
Resource Unit1 per
Evaluation2
Global Explanation4
500 Local Explanations5
Evaluations2 are calculations for
Generative AI or ML Quality
LLM or ML Drift
LLM or ML Model Health
ML Fairness
Predictive and Foundational model evaluations, monitoring, lifecycle tracking, and automatic documentation of facts.
Maximum 200 Resource Units1
Maximum 1k records/evaluation2
Maximum to 3 rows/use case
Maximum to 3 use cases
Maximum 500 local explanations per global explanation4
Maximum of 1 inventory
Predictive and Foundational model evaluations, monitoring, lifecycle tracking, and automatic documentation of facts.
Maximum 50k records /evaluation2
Maximum 500 local explanations per global explanation4
Maximum 500 inventories
USD 0.60 / Resource Unit
Future availability3
*Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
Watsonx.ai is available as software today. Pricing for watsonx.ai is based on the number of virtual processor cores (VPC). Software capabilities include Essentials SaaS tier.
Watsonx.data is available as software today. Pricing for watsonx.data software is based on the number of virtual processor cores (VPC). Software capabilities include Standard SaaS tier plus bundled IBM Analytics Engine Powered by Apache Spark, IBM Storage Fusion, IBM Storage Ceph, IBM Cloud Pak for Data platform software, Red Hat OpenShift and more.
Watsonx.governance is available as software today. Pricing for watsonx.governance software is based on the number of virtual processor cores (VPC). Software capabilities include Essentials SaaS tier.
Footnotes
1 Capacity Unit Hour pricing depends on the environment and tools utilized within a billing month.
2 For foundation model inference, charges are based on a Resource Unit (RU) metric which is equivalent to 1000 token (including both input and output tokens).
3 IBM’s statements regarding its plans, directions, and intent are subject to change or withdrawal without notice at IBM’s sole discretion. Information regarding potential future products is intended to outline our general product direction and it should not be relied on in making a purchasing decision. The information mentioned regarding potential future products is not a commitment, promise, or legal obligation to deliver any material, code or functionality. Information about potential future products may not be incorporated into any contract. The development, release, and timing of any future features or functionality described for our products remains at our sole discretion.
4 Global Explanation identifies features that have the most impact on the behavior of your model.
5 Local Explanation explains an individual transaction of a predictive ML model.
Unless otherwise specified under Software pricing, all features, capabilities, and potential updates refer exclusively to SaaS. IBM makes no representation that SaaS and software features and capabilities will be the same.
SaaS tiers apply to each platform component individually.