Explore the pricing tiers for our trial, essentials and standard plans on IBM® watsonx.ai™. For model pricing, explore IBM's foundation and embedding model section, as well as third-party foundation and embedding models pricing.
Charge metrics
Model inference based on resource unit (per 1000 tokens)2
ML functionality based on capacity unit-hour (per tool or runtime)1
Text extraction based on page
Model hosting based on hour
ML functionality (20 CUH limit per month)1
Inferencing (50,000 tokens per month)2
Prompt Lab
Open source models
IBM-developed IBM watsonx™ models
Synthetic Data Generator
ML functionality
Inferencing
Prompt Lab
Open source models
IBM-developed watsonx models
Synthetic Data Generator
Prompt tuning
Text extraction
USD 1050 per month* - tier fee
ML functionality
Inferencing
Prompt Lab
Open source models
IBM-developed watsonx models
Synthetic Data Generator
Prompt tuning
Text extraction
Custom foundation models
Model hosting
Deploy on demand models
*Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
granite-3-8b-instruct (v3.1)
USD 0.20*
granite-7b-lab
USD 5.22*
granite-3-8b-base
USD 5.22*
granite-3-2b-instruct (v3.1)
USD 0.10*
granite-guardian-3-8b (v3.1)
USD 0.20*
granite-guardian-3-2b (v3.1)
USD 0.10*
granite-20b-multilingual
USD 0.60*
granite-13b-instruct
USD 0.60*
USD 5.22*
granite-34b-code-instruct
USD 0.60*
granite-20b-code-instruct
USD 0.60*
granite-8b-code-instruct
USD 0.60*
granite-3b-code-instruct
USD 0.60*
granite-8b-japanese
USD 0.60*
granite-20b-code-base-schema-linking
USD 5.22*
granite-20b-code-base-sql-gen
USD 5.22*
*Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
granite-embedding-107m-multilingual
USD 0.10*
granite-embedding-278m-multilingual
USD 0.10*
slate-125m-english-rtrvr-v2
USD 0.10*
slate-125m-english-rtrvr
USD 0.10*
slate-30m-english-rtrvr-v2
USD 0.10*
slate-30m-english-rtrvr
USD 0.10*
*Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
deepseek-r1-distill-llama-70b
DeepSeek
USD 20.85*
deepseek-r1-distill-llama-8b
DeepSeek
USD 5.22*
llama-3-3-70b-instruct-hf
Meta
USD 20.85*
llama-3-2-90b-vision-instruct
Meta
USD 2.00*
llama-3-2-11b-vision-instruct
Meta
USD 0.35*
llama-guard-3-11b-vision
Meta
USD 0.35*
llama-3-2-1b-instruct
Meta
USD 0.10*
llama-3-2-3b-instruct
Meta
USD 0.15*
llama-3-405b-instruct
Meta
Input: USD 5.00*/ Output: USD 16.00*
llama-3-1-70b-instruct
Meta
USD 1.80*
USD 20.85*
llama-3-1-8b-instruct
Meta
USD 0.60*
USD 5.22*
llama-3-1-8b
Meta
USD 5.22*
llama-3-70b-instruct
Meta
USD 1.80*
USD 20.85*
llama-2-70b-chat
Meta
USD 20.85*
llama-2-13b-chat
Meta
USD 5.22*
allam-1-13b-instruct
SDAIA
USD 1.80*
codellama-34b-instruct
Meta
USD 1.80*
pixtral-12b
Mistral AI
USD 0.35*
mistral-large-instruct-2411
Mistral AI
USD 20.85**
mistral-large-2
Mistral AI
Input: USD 3.00*/ Output: USD 10.00*
Mistral-Small-24B-Instruct-2501
Mistral AI
USD 0.35*
mixtral-8x7b-instruct
Mistral AI
USD 0.60*
USD 10.40*
mixtral-8x7b-base
Mistral AI
USD 10.40*
mistral-nemo-instruct-2407
Mistral AI
USD 5.22*
jais-13b-chat (Arabic)
core42
USD 1.80*
flan-t5-xl-3b
USD 0.60*
USD 5.22*
flan-t5-xxl-11b
USD 1.80*
USD 5.22*
flan-ul2-20b
USD 5.00*
USD 5.22*
elyza-japanese-llama-2-7b-instruct
ELYZA
USD 1.80*
mt0-xxl-13b
BigScience
USD 1.80*
*Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
all-mini-l6-v2
Microsoft
USD 0.10*
all-minilm-l12-v2
OS-NLP-CV
USD 0.10*
multilingual-e5-large
Intel
USD 0.10*
*Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
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).
**Mistral commercial models have a GPU hosting fee and a model access fee. For more information, view the documentation
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.