Granite 4.2 8B

Open Source

Open Weights

Parameters

Context Length

131K

Modality

Reasoning

Architecture

Dense

License

Apache 2.0

Release Date

7 Aug 2026

Knowledge Cutoff

API Pricing (per 1M)

Input: $0.06 · Output: $0.25

System Requirements

VRAM requirements for different quantization methods and context sizes

1,024 tokens

17.31 GB VRAM

Consumer

1x RTX 4090

24GB VRAM

Datacenter

1x NVIDIA A100

80GB VRAM

Apple Silicon

1x Apple M3 Max

128GB VRAM

131,072 tokens

38.15 GB VRAM

Consumer

2x RTX 4090

24GB VRAM

Datacenter

1x NVIDIA A100

80GB VRAM

Apple Silicon

1x Apple M3 Max

128GB VRAM

Architecture Diagram

Evaluation Benchmarks

Rank

#149

Benchmark	Score	Rank
Professional Knowledge MMLU Pro	0.74	37
Graduate-Level QA GPQA	0.641	94
Agentic Index Artificial Analysis	0.01	119
Coding Index Artificial Analysis	0.22	129
General Text Text Arena	1290	157
Intelligence Index Artificial Analysis	0.11	249

Rankings

Overall Rank

#149

Coding Rank

#134

About Granite 4.2 8B

Granite 4.2 8B is a dense reasoning model developed by IBM for enterprise and multi-step reasoning applications. It is tailored for mathematics, code generation, multilingual dialogue, and agentic workflows.

Technical Specifications

Attention

Attention Structure

Grouped-Query Attention

Attention Heads

Key-Value Heads

Attention Head Dimension

128

Position Embedding

ROPE

RoPE Theta

10,000,000

Sliding Window Attention

Sliding Window Size

Sliding Window Ratio

Linear Attention

Linear Attention Ratio

Normalization

RMS Normalization

Activation Function

SwigLU

Dimensions

Auxiliary Parameters

Hidden Dimension Size

4,096

Number of Layers

FFN Intermediate Size (Dense)

12,800

Multi-Token Prediction Heads

Tokenizer

Vocabulary Size

100,352

Resources

Download Weights

About Granite

The Granite model family developed by IBM.

Other Granite Models

No related models available