Granite 4.2 30B
Granite 4.2 30B is a dense text model post-trained from Granite 4.1 30B Base. It supports full thinking, direct answers and a low-effort reasoning mode. IBM released it on 25 August 2026.
Model design and training
Training combines supervised fine-tuning, alignment and reinforcement learning. The 30B training recipe adds a second supervised phase that increases the share of agent traces. The chat template controls thinking behavior.Granite 4.2 30B model card
Practical use
Practical guidance. The 30B model targets more demanding reasoning and tool workflows. Its dense design requires memory for the full model during inference. Assess the complete chain of tool decisions and final answers. A higher score on an isolated question does not establish reliability across a long agent run.
Context and serving limits
The card advertises a 512K extension, but its configuration and serving examples use 131,072 tokens. Treat extended context as a separate deployment setting to verify. The 512K claim is not used here as the default endpoint limit.Granite 4.2 30B model card
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | Downloadable |
| Model identifiers | ibm-granite/granite-4.2-30b |
| Context window | 131,072 tokens Default published configuration and architecture table. IBM also describes extension to 512K. Granite 4.2 30B model card |
| License | Apache 2.0Granite 4.2 30B model card |
| Parameters | 30BGranite 4.2 30B model card |
| Supported languages | 12 languages, including English, German, French, Chinese and Japanese.Granite 4.2 30B model card |
Related models
- Granite 4.1 30B
IBM · Granite 4.1
An instruction-tuned 30B text model for retrieval-based answers, extraction, coding and tool use.
Launch date API · Downloadable · Provider product - Granite 4.1 30B Base
IBM · Granite 4.1
A pretrained 30B dense decoder for text generation, further training and fill-in-the-middle code completion.
Launch date Downloadable