A one-stop, integrated AI development studio for end-to-end AI application development
watsonx.ai pricing
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.
Includes pay-as-you-go pricing per million tokens and hourly rates for on-demand model hosting and deployment.
Includes IBM and third-party models available for USD 0.10 per million tokens.
Includes third-party models from Meta, Google, DeepSeek, Mistral, and more, with pay-as-you-go pricing per million tokens and hourly options for on-demand hosting and deployment.
Includes use case-based pricing for machine learning, text extraction, and model customization, with Essential and Standard package options.
Pricing plans — multiple payment options available
Foundation Models: Up to 300,000 tokens per month
Machine Learning Tools: Up to 20 Compute Usage Hours (CUH) per month
Text Extraction: Up to 100 documents per month
Playground UI
Inferencing
Open source models
IBM watsonx® models
Work with foundational models (PromptLab)
Supports retrieval augmented generation (RAG)
Work with agents (AgentLab)
Synthetic data generator
ML functionality**
Text extraction**
LoRA/QLoRA Fine-tuning*
Custom foundation models***
Model hosting***
Deploy on-demand models***
Support
watsonx community and online chatbot
Basic support included: 24x7 access to tech support through cases
Basic support included: 24x7 access to tech support through cases
Options available
Advanced support with SLAs available starting at USD 200 per month
Advanced support with SLAs available starting at USD 200 per month
*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-speech-4-1-2b
IBM
Not available
granite-4-1-3b
IBM
Not available
granite-4-1-8b
IBM
Not available
granite-4-1-30b
IBM
Not available
granite-vision-4-1-4b
IBM
Not available
granite-4h-micro
IBM
Not available
granite-4h-tiny
IBM
Not available
granite-4h-small
IBM
USD 0.0636 per 1M tokens input
USD 0.265 per 1M tokens output
granite-vision-3-3-2b
IBM
Not available
granite-3-3-2b-instruct
IBM
Not available
granite-3-3-8b-instruct
IBM
Not available
granite-3-2-8b-instruct
IBM
Not available
granite-3-1-8b-base
IBM
Not available
granite-3-1-8b-instruct
IBM
Not available
granite-3-8b-base
IBM
Not available
granite-13b-chat-v2
IBM
Not available
granite-3b-code-instruct
IBM
Not available
granite-8b-code-instruct
IBM
USD 0.636
granite-20b-code-base-sql-gen
IBM
Not available
granite-20b-code-instruct
IBM
Not available
granite-20b-multilingual
IBM
Not available
granite-20b-code-base-schema-linking
IBM
Not available
granite-34b-code-instruct
IBM
Not available
granite-7b-lab
IBM
Not available
granite-8b-japanese
IBM
Not available
granite-guardian-3-8b
IBM
USD 0.212
*Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
Meta models
llama-4-maverick-17b-128e-instruct-int4
Meta
Not available
llama-4-maverick-17b-128e-instruct-fp8
Meta
USD 0.371 per 1M tokens input
USD 1.484 per 1M tokens output
lama-4-scout-17b-16e-instruct-fp8-dynamic
Meta
Not available
llama-3-3-70b-instruct
Meta
USD 0.7526
llama-3-3-70b-instruct-hf
Meta
Not available
llama-3-2-11b-vision-instruct
Meta
USD 0.371
llama-3-2-90b-vision-instruct
Meta
Not available
llama-3-1-405b-instruct-fp8
Meta
Not available
llama-3-1-8b
Meta
Not available
llama-3-1-8b-instruct
Meta
Not available
llama-3-1-70b
Meta
Not available
llama-3-1-70b-gptq
Meta
Not available
llama-3-1-70b-instruct
Meta
Not available
llama-3-1-405b-instruct-fp8
Meta
Not available
llama-3-8b-instruct
Meta
Not available
llama-3-70b-instruct
Meta
Not available
llama-2-70b-chat
Meta
Not available
codellama-34b-instruct-hf
Code Llama
Not available
llama-guard-3-11b-vision
Meta
USD 0.371
Not available
*Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
Mistral models
mistral-medium-3-5-0
Mistral AI
Not available
devstral-small-2512
Mistral AI
Not available
mistral-large-2512
Mistral AI
USD 0.636 per 1M tokens input
USD 1.908 per 1M tokens output
ministral-14b-instruct-2512
Mistral AI
Not available
ministral-8b-instruct-2512
Mistral AI
Not available
ministral-3b-instruct-2512
Mistral AI
Not available
mistral-medium-2508
Mistral AI
Not available
devstral-medium-2507
Mistral AI
Not available
devstral-small-2507
Mistral AI
Not available
mistral-small-3-2-24b-instruct-2506
Mistral AI
Not available
mistral-medium-2505
Mistral AI
USD 3.18 per 1M tokens input
USD 9.5per 1M tokens output
mistral-small-3-1-24b-instruct-2503
Mistral AI
USD 0.106 per 1M tokens input
USD 0.318 per 1M tokens output
codestral-2501
Mistral AI
Not available
mistral-large-instruct-2411
Mistral AI
Not available
ministral-8b-instruct-2410
Mistral AI
Not available
mistral-large-instruct-2407
Mistral AI
Not available
mistral-nemo-instruct-2407
Mistral AI
Not available
mixtral-8x7b-base
Mistral AI
Not available
mixtral-8x7b-instruct-v01
Mistral AI
Not available
pixtral-12b
Mistral AI
Not available
*Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
nvidia-nemotron-3-ultra-550b-a55b-quantized
Red Hat
Not available
gpt-oss-120b
Open AI
USD 0.159 per 1M tokens input
USD 0.636 per 1M tokens output
gpt-oss-20b
Open AI
Not available
deepseek-r1-distill-llama-8b
DeepSeek
Not available
deepseek-r1-distill-llama-70b
DeepSeek
Not available
allam-1-13b-instruct
SDAIA
Not available
llama-3-1-nemotron-ultra-253b-v1-fp8
NVIDIA
Not available
nvidia-nemotron-3-super-120b-a12b-fp8
NVIDIA
Not available
nvidia-nemotron-nano-12b-v2-vl-fp8
NVIDIA
Not available
eurollm-1-7b-instruct
Utter Project
Not available
eurollm-9b-instruct
Utter Project
Not available
mt0-xxl-13b
BigScience
Not available
poro-34b-chat
LumiOpen
Not available
*Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
Embedding model library
granite-embedding-278m-multilingual
USD 0.106 per million tokens
Not available
slate-125m-english-rtrvr-v2
USD 0.106 per million tokens
Not available
slate-30m-english-rtrvr-v2
USD 0.106 per million tokens
Not available
*Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
Third-party embedding 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.
Machine learning models
0.55 USD / Capacity Unit-Hour
0.45 USD / Capacity Unit-Hour
Text extraction3
0.0403 USD / Page
0.0318 USD / Page
LoRA fine-tuning
Not available
NVIDIA 1 x A100 GPU: 6.3 USD / Hour
NVIDIA 1 x H100 GPU: 14.85 USD / Hour
Model hosting/Deploy on demand
Not available
NVIDIA 1 x L40S GPU: 4.43 USD / Hour
NVIDIA 2 x L40S GPU: 8.86 USD / Hour
NVIDIA 1 x A100 GPU: 5.8 USD / Hour
NVIDIA 2 x A100 GPU: 11.6 USD / Hour
NVIDIA 4 x A100 GPU: 23.2 USD / Hour
NVIDIA 8 x A100 GPU: 46.4 USD / Hour
NVIDIA 1 x H100 GPU: 14.5 USD / Hour
NVIDIA 2 x H100 GPU: 29 USD / Hour
NVIDIA 4 x H100 GPU: 58 USD / Hour
NVIDIA 8 x H100 GPU: 116 USD / Hour
NVIDIA 1 x H200 GPU: 16 USD / Hour
NVIDIA 2 x H200 GPU: 32 USD / Hour
NVIDIA 4 x H200 GPU: 64 USD / Hour
NVIDIA 8 x H200 GPU: 128 USD / Hour
*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 For foundation model inference, charges are based on a Resource Unit (RU) metric equivalent to 1000 tokens (including both input and output tokens).
2 Mistral commercial models have a GPU hosting fee and a model access fee. For more information, view the documentation.
* Prices shown are indicative, may vary by country, exclude any applicable taxes and duties, and are subject to product offering availability in a locale.
^ Capacity Unit Hour pricing depends on the environment and tools utilized within a billing month.
3 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.