IBM: Granite 4.1 8B logo

MODELS

IBM: Granite 4.1 8B

by IBM

modellead-sourceopenrouter-modelslanguage-modelllmibmgranitegranite-4.18blong-contextenterpriseapache-2.0openrouterhugging-facesource:research.ibm.cominstructwatsonx-ai
Visit WebsiteDocumentation

Overview

IBM Granite 4.1 8B is an 8B-parameter long-context instruct language model in IBM’s Granite 4.1 family for general-purpose enterprise applications.

Details

IBM Granite 4.1 8B is part of IBM’s Granite 4.1 dense language-model family, which IBM documentation describes in 3B, 8B, and 30B sizes. The Hugging Face model card identifies Granite-4.1-8B as an 8B-parameter long-context instruct model developed by the Granite Team at IBM and released under Apache 2.0. IBM Research describes Granite 4.1 as dense decoder-only language models in base and instruct sizes, with the 8B instruct model targeting instruction following and tool calling for enterprise workloads. IBM’s watsonx.ai foundation-model page lists granite-4-1-8b as available with a 131K context length and use cases including summarization, extraction, RAG, code tasks, and function calling.

When to Use

Use for general-purpose enterprise language-model applications where an 8B IBM Granite model is appropriate. Use when you need a long-context instruct model from IBM’s Granite 4.1 family. Use for supported watsonx.ai use cases listed by IBM, including summarization, extraction, RAG, code tasks, and function calling. Use when Apache 2.0 licensing is a requirement for model evaluation or deployment.

Getting Started

  1. Review the IBM Granite 4.1 model documentation for the 3B, 8B, and 30B Granite 4.1 family.
  2. Open the IBM Granite Hugging Face model card for Granite-4.1-8B to confirm model identity, license, and usage details.
  3. Check IBM’s watsonx.ai foundation-model library to verify granite-4-1-8b availability, context length, and supported use cases.
  4. Use IBM watsonx.ai API documentation if you plan to run foundation-model text inference, chat completions, or model-listing workflows through IBM Cloud.
  5. Check IBM watsonx.ai pricing for current pay-as-you-go or deploy-on-demand availability status.

Key Features

  • •8B-parameter IBM Granite 4.1 instruct model
  • •Long-context model identified by IBM’s Hugging Face model card
  • •Part of IBM’s Granite 4.1 family of 3B, 8B, and 30B dense language models
  • •131K context length listed on IBM’s watsonx.ai foundation-model page
  • •Targets instruction following and tool calling for enterprise workloads according to IBM Research
  • •Apache 2.0 license

Capabilities

  • •Instruction-following language generation
  • •Long-context text processing
  • •Summarization
  • •Extraction
  • •Retrieval-augmented generation use cases
  • •Code tasks
  • •Function calling
  • •Foundation-model inference through IBM watsonx.ai APIs where available

Last updated May 31, 2026