Granite 4.0 1B
Granite 4.0 1B is a compact instruction-tuned text model. IBM reports 1.6B parameters for this checkpoint. It belongs to the Nano releases and is adapted from its matching base model.
Model design and training
The dense architecture uses rotary position embeddings and grouped-query attention. IBM applies supervised fine-tuning, reinforcement learning and model merging to the corresponding base. The resulting checkpoint follows instructions rather than serving only as a pretrained completion model.Granite 4.0 1B model card
Practical use
Practical guidance. The 1B name is a family label; plan memory from the published 1.6B parameter count. Use short representative tasks to establish output quality before supplying very long documents. A long context window does not ensure that a compact model finds every relevant fact.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | ibm-granite/granite-4.0-1b |
| Context window | 131,072 tokens Published model context; deployment settings can impose a smaller limit. Granite 4.0 1B model card |
| License | Apache 2.0Granite 4.0 1B model card |
| Parameters | 1.6BGranite 4.0 1B model card |
| Architecture | 40 attention layers; hidden size 2,048.Granite 4.0 1B model card |
Related models
- Granite 4.0 H-1B
IBM ยท Granite 4.0 Nano
A 1.5B-parameter hybrid instruction model in IBM's compact Granite Nano text family.
Launch date not confirmedDownloadable