Granite 4.1 8b is an open-source foundation model hosted on Hugging Face under the repository ibm-granite/granite-4.1-8b. It functions as a language model that processes text and supports structured interactions including tool use and document referencing.
The model includes a defined tokenizer with specific tokens such as a beginning-of-text marker, end-of-text marker, pad token, and unknown token. Its chat template is implemented in Jinja format and provides system messages for handling tools via XML-tagged function calls that return JSON objects containing a function name and arguments. A separate document-oriented system message prefix is also defined for scenarios involving provided documents. These elements enable the model to manage assistant-like interactions that incorporate external tools or reference materials when responding to queries.
It is delivered as a downloadable model on the Hugging Face platform, where users can access it alongside other models, datasets, and related resources. The surrounding platform supports community features, documentation, and inference options but the model entry itself centers on the architecture and template definitions for text-based tasks.
In the Foundation models & chat space, Granite 4.1 8b takes a focused approach. It focuses on providing an open-source large language model for developers to build advanced NLP solutions. Granite 4.1 8b is an open-source project aimed at AI researchers and developers. The project is open source (Open Source). It runs on the command line, and it can be self-hosted.
IBM builds and maintains Granite 4.1 8b, and the product first shipped in 2024. Among its 5 catalogued features are text generation, transformer architecture, and docker support.
Latest indexed changes and source events
ibm-granite/granite-4.1-8b verified by the PulseGate indexer