Granite GGUF models

1547 GGUF models in the Granite family, quantized by 403 uploaders. Sizes run from 2M to 33.7B parameters. The longest trained context is 5120K tokens.

TinySmallMediumLarge
Most downloadedNewestLargestLongest contextSmallest
GR
Granite 4.2 3b
Granite · ibm-granite
Params3.66B
Context128K
Q4_K_M2.2 GB
GGUF
GR
Granite 4.2 8b
Granite · ibm-granite
Params8.79B
Context128K
Q4_K_M5.3 GB
GGUF
GR
Granite 4.2 30b
Granite · ibm-granite
Params29.3B
Context128K
Q4_K_M17.7 GB
GGUF
GR
Parable Granite 4.1 3B Claude Fable 5
Granite · AnkitAI
Params3.4B
Context128K
Q4_K_M2.1 GB
ReasoningTool use
GR
Parable Granite 4.1 8B Claude Fable 5
Granite · AnkitAI
Params8.38B
Context128K
Q4_K_M5.1 GB
ReasoningTool use
GR
Granite 4.2 3B BE
Granite · taronaeo
Params3.66B
Context—
Q4_K_M2.2 GB
ReasoningTool use
GR
Granite 4.0 1B BE
Granite · taronaeo
Params1.63B
Context—
Q4_K_M1.0 GB
GGUF
GR
ibm granite granite 20b code instruct
Granite · RichardErkhov
Params1.9B
Context—
Q4_K_M0.0 GB
Coding
GR
ibm granite granite 20b code base
Granite · RichardErkhov
Params1.9B
Context—
Q4_K_M0.0 GB
Coding
GR
Granite Speech 4.1 2B NAR
Granite · handy-computer
Params2.25B
Context—
Q4_K_M1.6 GB
GGUF
GR
Granite Speech 4.1 2B
Granite · handy-computer
Params2.31B
Context—
Q4_K_M1.6 GB
GGUF
GR
Granite 4.0 350M BE
Granite · taronaeo
Params352M
Context—
Q4_K_M0.2 GB
GGUF
GR
Granite 4.0 H 350M BE
Granite · taronaeo
Params340M
Context—
Q4_K_M0.2 GB
GGUF
GR
Granite Embedding 107m Multilingual
Granite · bartowski
Params107M
Context512
Q4_K_M0.1 GB
Embedding
GR
Granite-4.1-3B
Granite · weareapexcreators
Params3.4B
Context128K
Q4_K_M2.1 GB
GGUF
GR
granite 4.0 1b speech
Granite · handy-computer
Params2.31B
Context—
Q4_K_M1.6 GB
GGUF
GR
Granite Speech 4.1 2B Plus
Granite · handy-computer
Params2.11B
Context—
Q4_K_M1.5 GB
GGUF
GR
Granite 4.1 8b
Granite · ibm-granite
Params8.79B
Context128K
Q4_K_M5.3 GB
GGUF
GR
TwIL LM3 Pro
Granite · webAI-Official
Params3.66B
Context128K
Q4_K_M2.2 GB
ReasoningTool use
GR
Granite 4.1 3b
Granite · ibm-granite
Params3.4B
Context128K
Q4_K_M2.1 GB
GGUF
GR
Granite 20b Code Instruct
Granite · bartowski
Params20B
Context8K
Q4_K_M12.8 GB
CodingTool use
GR
Granite 4.0 H Tiny
Granite · ibm-granite
Params6.94B
Context1024K
Q4_K_M4.2 GB
GGUF
GR
Granite 4.0 H Micro
Granite · ibm-granite
Params3.2B
Context1024K
Q4_K_M1.9 GB
GGUF
GR
Granite 4.0 Micro
Granite · ibm-granite
Params3.4B
Context128K
Q4_K_M2.1 GB
GGUF
GR
Granite Vision 4.1 4b
Granite · ibm-granite
Params3.4B
Context128K
Q4_K_M2.1 GB
Vision
GR
Granite-4.0-H-Micro
Granite · unsloth
Params3.2B
Context1024K
Q4_K_M1.9 GB
GGUF
GR
Granite-4.0-350M
Granite · unsloth
Params352M
Context32K
Q4_K_M0.2 GB
GGUF
GR
Granite 3.3 8b Instruct
Granite · ibm-granite
Params8.17B
Context128K
Q4_K_M4.9 GB
GGUF
GR
Granite-4.0-H-Tiny
Granite · unsloth
Params6.94B
Context1024K
Q4_K_M4.3 GB
GGUF
GR
Granite 4.0 H Small
Granite · ibm-granite
Params32.2B
Context1024K
Q4_K_M19.5 GB
GGUF

About the Granite GGUF models

Granite is represented on ggufmodels.com by 1,547 GGUF models, quantized by 403 uploaders on Hugging Face. The most downloaded are Granite 4.2 3b, Granite 4.2 8b and Granite 4.2 30b. Parameter counts run from 2M to 33.7B, and the longest trained context window is 5120K tokens.

A Q4_K_M file of a Granite model is between 0.0 to 21.4 GB, which is the quant most people start with. Most files come from mradermacher (367), mrutkows (62) and ibm-granite (48). The quants you will meet most often are Q4_K_M (1100), Q8_0 (880), Q5_K_M (872) and Q6_K (832), where the number is how many models offer that quant.

Frequently asked questions

Which Granite GGUF should I download?

The most downloaded Granite GGUF here is Granite 4.2 3b. Pick the size that fits your GPU first, then choose a quant: Q4_K_M is the usual starting point, and Q5_K_M or Q6_K give more quality if you have memory to spare. The VRAM pages below list what fits each GPU.

How much VRAM does a Granite model need?

Q4_K_M files of Granite models on this site range from 0.0 to 21.4 GB. Add the KV cache for your context length and about 0.5 GB for the runtime, which usually means 1.5 to 3 GB on top of the file at an 8K context.

Where do Granite GGUF files come from?

The most common quantizers here are mradermacher, mrutkows and ibm-granite. They convert the original model and publish quantized files on Hugging Face. ggufmodels.com does not host files; every download button opens Hugging Face.

Keep browsing