Sold by: IBM Data and AI
Deployed on AWS
IBM Granite 3.2 Instruct is an open-source model with controllable reasoning, offering strong performance and enhanced complex thinking.
Overview
IBM Granite 3.2 Instruct is a family of 2B and 8B parameter language models fine-tuned for enhanced reasoning capabilities. Built on Granite 3.1, it uses permissively licensed open-source datasets and synthetic data optimized for reasoning tasks. A key feature is its controllable thinking capability, which can be toggled on or off to optimize computational efficiency. Released under Apache 2.0, it supports 12 languages, including English, German, Spanish, French, Japanese, and Chinese, with extensibility for additional languages. Unlike industry trends that separate reasoning models, IBM integrates reasoning directly into the core Instruct models. While traditional approaches improve logic-based tasks at the cost of others, IBM’s method enhances reasoning without trade-offs. It excels in summarization, classification, extraction, QA, RAG, code tasks, function-calling, and multilingual dialogues, performing strongly on prominent benchmarks without sacrificing other capabilities.
Highlights
- IBM Granite 3.2 introduces controllable reasoning capabilities that can be toggled on or off with a simple parameter, allowing developers to balance computational efficiency with enhanced problem-solving. This unique approach preserves general performance while significantly improving complex instruction following.
- Unlike competing reasoning models that sacrifice general capabilities for narrow domains, Granite 3.2 demonstrates substantial improvements on benchmarks like ArenaHard and AlpacaEval without compromising performance elsewhere, maintaining IBM's commitment to safety and comprehensive functionality.
- Granite 3.2 applies IBM's Thought Preference Optimization framework to enhance reasoning without the extensive computation typically required by other models. This practical approach delivers enterprise-ready performance across summarization, classification, RAG, code tasks, and multilingual support in 12 languages.
Details
Sold by
Delivery method
Latest version
Deployed on AWS
New
Introducing multi-product solutions
You can now purchase comprehensive solutions tailored to use cases and industries.
Features and programs
Financing for AWS Marketplace purchases
AWS Marketplace now accepts line of credit payments through the PNC Vendor Finance program. This program is available to select AWS customers in the US, excluding NV, NC, ND, TN, & VT.
Pricing
This product is available free of charge. Free subscriptions have no end date and may be canceled any time.
Additional AWS infrastructure costs may apply. Use the AWS Pricing Calculator to estimate your infrastructure costs.
Dimensions summary
This listing is free to use; you pay only for the AWS instance hours you run. Pricing splits into two inference modes. Batch mode runs on two instance types (ml.g5.12xlarge and ml.g5.24xlarge) for processing grouped requests. Real-time mode covers the remaining instance types (g6e, g6, p4d, and p5 families) for on-demand responses. Within each mode, you pick an instance size based on the compute and memory you need. Larger instance sizes carry higher hourly rates. Billing is per host hour, so cost scales with how long each instance runs.
Top-of-mind questions for buyers
One host hour is one hour that a chosen instance runs the model. The software itself is free, so you pay only AWS instance charges. A fully stopped instance stops accruing host-hour charges. Stored data or attached volumes may still carry separate AWS storage fees while stopped.
Both modes bill per host hour. Batch mode processes grouped requests together and runs on the ml.g5.12xlarge and ml.g5.24xlarge instances. Real-time mode answers on-demand requests and runs on the g6e, g6, p4d, and p5 instances. Batch suits scheduled bulk jobs; real-time suits live, interactive responses.
Each instance type offers a different mix of compute and memory. Larger sizes handle heavier request volumes and carry higher hourly rates. You pick one instance type per endpoint. The model supports tool calling, code generation, and instruction following, which shape the throughput you need.
www.ibm.com
Helpful?
Vendor refund policy
This product is offered for free. If there are any questions, please contact us for further clarifications.
Legal
Vendor terms and conditions
Upon subscribing to this product, you must acknowledge and agree to the terms and conditions outlined in the vendor's End User License Agreement (EULA) .
Content disclaimer
Vendors are responsible for their product descriptions and other product content. AWS does not warrant that vendors' product descriptions or other product content are accurate, complete, reliable, current, or error-free.
Delivery details
Amazon SageMaker model
An Amazon SageMaker model package is a pre-trained machine learning model ready to use without additional training. Use the model package to create a model on Amazon SageMaker for real-time inference or batch processing. Amazon SageMaker is a fully managed platform for building, training, and deploying machine learning models at scale.
Deploy the model on Amazon SageMaker AI using the following options:
Version release notes
IBM Granite 3.2 Instruct is a reasoning-enhanced model in 2B and 8B sizes, built on Granite 3.1. Trained on permissive and synthetic reasoning data, it allows developers to toggle its reasoning process on and off via a simple parameter. Unlike other models, Granite 3.2 improves complex instruction following without sacrificing general performance. It excels in summarization, classification, QA, RAG, coding, and function calling. Supporting 12 languages, it is ideal for enterprise use where strong reasoning and efficiency are key.
Additional details
Inputs
- Summary
The model can be invoked by passing a prompt. Please see the sample notebook for details.
- Input MIME type
- application/json
https://github.com/ibm-granite-community/SageMaker/blob/main/granite-3.2-8b-instruct/real_time_sample_input_data.json
https://github.com/ibm-granite-community/SageMaker/blob/main/granite-3.2-8b-instruct/batch_sample_input_data.json
Input data descriptions
The following table describes supported input data fields for real-time inference and batch transform.
Field name | Description | Constraints | Required |
|---|---|---|---|
inputs | The prompt to be passed to the model. | - | Yes |
Resources
Support
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.
Similar products
IBM's Granite 3.0 8B Instruct is an 8B-parameter AI model for enterprise use, excelling in multilingual and code tasks; Apache 2.0 licensed.
IBM Granite 3.2 Instruct is an open-source model with controllable reasoning, offering strong performance and enhanced complex thinking.
IBM Granite 3.1 8B is an open-source model with a 128K token context length, optimized for enterprise AI and multilingual tasks.
IBM Granite 3.3 Instruct 8B is an open-source model with enhanced reasoning, multilingual support, and structured output formatting.
Customer reviews
No customer reviews yet
Be the first to review this product . We've partnered with PeerSpot to gather customer feedback. You can share your experience by writing or recording a review, or scheduling a call with a PeerSpot analyst.