IBM watsonx Plans

Access is provided through the watsonx.ai playground and Prompt Lab rather than as unrestricted access to IBM's entire model catalog. The exact available model set can vary by region, offering availability, and current IBM catalog configuration. Foundation-model use is subject to the plan's monthly token allowance.

Foundation models are limited to up to 300,000 tokens per month. Machine-learning tools are limited to up to 20 Compute Usage Hours per month. Text extraction is limited to up to 100 documents per month. The underlying free watsonx.ai services use Lite-level resources.

Not intended for production deployments or enterprise-scale capacity. Usage is capped by monthly token, compute-hour, and document allowances. Advanced production functions such as model hosting, on-demand deployment, custom foundation models, and LoRA/QLoRA fine-tuning are not included in the free playground offering.

Free Toolbox playground. IBM advertises up to 300,000 foundation-model tokens per month, up to 20 Compute Usage Hours per month, and up to 100 extracted documents per month. Prices and availability may vary by country.

Essentials includes inferencing and access to supported IBM watsonx and open-source models, but model availability and pricing are model-specific and region-dependent. Foundation-model inference is metered separately; the plan does not provide unlimited model usage. On-demand model hosting is reserved for Standard.

Usage is consumption-based rather than a fixed allowance. Foundation-model inference is billed using Resource Units, with one Resource Unit equivalent to 1,000 input and output tokens. Machine-learning and text-extraction usage are separately metered. The Essentials runtime documentation describes usage as CUH-based and lists a 3-day deployment-to-idle period and up to 5 parallel Decision Optimization batch jobs per deployment.

LoRA/QLoRA fine-tuning and on-demand model hosting/deployment are not available in the Essentials pricing tier. Custom foundation models are associated with the Standard tier. Essentials does not include the Standard plan's fixed 2,500-CUH monthly block or enterprise production capacity. Model and feature availability can vary by region.

IBM advertises Essentials as starting at USD 0/month, but usage is metered. Current listed examples include foundation-model inference priced by resource units, machine-learning models at USD 0.55 per Capacity Unit-Hour, and text extraction at USD 0.0403 per page. Prices may vary by country and exclude taxes. Advanced support with SLAs is available from USD 200/month.

IBM documents generative AI and conversational-search capabilities for Plus, but does not provide a stable plan-specific public list of exact model identifiers. Model selection and availability may vary by region and service configuration.

Billing is based on monthly active users rather than seats or API calls. IBM documents unlimited monthly active users for Plus. The plan supports up to 100 concurrent phone calls and includes plan-specific resource limits such as 10 assistants, 50 skills, 100 integrations, 1,000 entities, 2,000 intents, and 30 days of logs.

Phone integration and voice usage incur additional charges. Conversational search can incur add-on charges. Enterprise-only features include activity tracking, multiple environments, response-mode overrides, and Segment event sending. IBM is rolling out automatic upgrades of eligible Plus instances to watsonx Orchestrate Essentials.

IBM documentation states that Plus starts at USD 140 per month and includes the first 1,000 monthly active users. Additional active users are charged at USD 14 per 100 users per month. Phone integration voice usage is an additional USD 9 per 100 users per month. IBM's current catalog is transitioning watsonx Assistant instances to watsonx Orchestrate, so pricing availability may depend on account eligibility and region.

The plan includes core AI and LLM capabilities, but IBM's current public plan comparison does not identify a fixed list of exact model IDs for Essentials. Available foundation models and inference usage may depend on the watsonx Orchestrate deployment and applicable IBM Cloud entitlements.

Includes 60,000 skill runs per month and 4,000 monthly active users according to IBM licensing documentation. IBM's pricing comparison also lists up to 200,000 messages per month, up to 10 GB of documents for agent knowledge, and capped message usage per user. Unused skill runs do not roll over.

Usage is capped by monthly skill-run, active-user, message, document, and concurrency entitlements. Data isolation, dedicated data space, HIPAA-ready options, and enterprise-grade scale are Premium features. Web research and other partner capabilities may require optional add-ons. Pricing and feature availability can vary by deployment environment and region.

Starting price shown by IBM for the United States. Prices may vary by country, exclude taxes and duties, and depend on product availability. Additional usage and optional add-ons may incur separate charges.

IBM's current watsonx.ai pricing page identifies access to IBM and third-party foundation-model categories, but it does not provide a definitive plan-specific fixed model list for Standard. Model access and cost depend on the selected model, inference method, region, and whether the model is hosted or deployed on demand.

The plan is designed for high-capacity enterprise use. IBM watsonx.ai Runtime Standard includes 2,500 capacity unit hours per month, with additional usage billed on a pay-as-you-go basis. Other capabilities may use resource units, capacity unit hours, per-page pricing, token pricing, or hourly GPU/model-hosting rates.

The plan is not a flat unlimited-use subscription. Model inference, compute, text extraction, fine-tuning, model hosting, and on-demand deployment can incur usage-based charges. Availability of features depends on region and deployment environment. The legacy watsonx.ai Studio Standard plan is no longer available for new accounts; IBM documentation identifies the current paid Studio plan as Professional, while the current watsonx.ai product pricing page separately lists a Standard production plan.

IBM advertises the Standard plan as starting at USD 1,110 per month. Prices are indicative, may vary by country, exclude taxes and duties, and are subject to local availability. Additional model, compute, text-extraction, fine-tuning, hosting, and deployment usage can be metered separately. The Runtime Standard plan includes a block of 2,500 capacity unit hours per month; usage above the included amount is billed at the Standard rate.

Standard includes the core AI and LLM capabilities listed for Essentials and access to Premier LLM capability. IBM documentation indicates that selecting a Premier model incurs additional monthly active-user consumption beyond standard included usage. Exact model availability depends on deployment, region, model configuration, and IBM's current supported-model list.

The IBM Cloud Standard entitlement includes up to 450,000 skill runs per month and 40,000 monthly active users. The pricing comparison lists up to 2,000,000 messages per month, up to 10 GB of documents for agent knowledge, capped message usage per user, and higher capacity ceilings than Essentials. Unused skill runs do not roll over.

Pricing is a starting price and may vary by country and deployment. Usage is subject to monthly skill-run, active-user, message, document, concurrency, and capacity limits. Data isolation and dedicated data space are Premium features. Web research through the Bright Data Web Agent is an optional add-on rather than a standard-plan inclusion. Premier model usage may consume additional monthly active users.

IBM advertises Standard pricing starting at 6,360 USD per month. Prices are indicative, may vary by country, exclude taxes and duties, and depend on product availability in the locale.

IBM's current Enterprise plan documentation describes assistant capabilities and service limits but does not enumerate specific current model IDs or model families included with this exact plan.

The web chat documentation lists Enterprise as supporting unlimited monthly active users. Billing for Enterprise is based on monthly active users rather than API calls. Exact capacity limits for other resources are not publicly specified in the reviewed current documentation.

The Enterprise plan is distinct from Enterprise with Data Isolation. Enterprise with Data Isolation provides a dedicated environment and additional isolation-related capabilities and must be provisioned separately rather than upgraded in place. IBM is rolling out automatic upgrades of eligible watsonx Assistant instances to watsonx Orchestrate, formerly Assistant, so product access and naming may transition over time.

IBM does not publish a fixed public price for the Enterprise plan in the current plan documentation. Enterprise and Plus usage is measured by monthly active users, and enterprise arrangements may require IBM Cloud billing or contacting IBM. An Enterprise with Data Isolation variant is provisioned separately.

Premium inherits Standard capabilities, including access to IBM and third-party foundation-model functionality used by watsonx Orchestrate. IBM documents model access at the service level rather than publishing a Premium-specific fixed model list; available models and model classes may change at IBM's discretion.

The public pricing comparison lists monthly active users, messages per month, documents for agent knowledge, channels, and message usage as Custom for Premium. IBM Cloud entitlement variants include up to 50,000 monthly active users for Premium Agentic monthly active users and up to 500,000 messages per month for Premium Agentic messages. AWS Agentic Premium provides 7,500 resource units per month; unused resource units do not roll over.

Pricing, capacity, and entitlements are negotiated and may vary by deployment and contract. Optional partner agents and add-ons are not automatically included. IBM states that existing watsonx Orchestrate instances cannot be upgraded in place to Premium with data isolation; customers must create a new Premium tenant and migrate the environment. AWS data isolation availability is documented as limited to HIPAA-compliant clusters, with commercial AWS support subject to availability.

IBM lists Premium as Contact us rather than publishing a fixed public price. Pricing and availability may vary by country, deployment environment, configuration, and applicable taxes or duties.