IBM

Granite 4.2 8B

IBM·8B·Granite·128K ctx·2026-08

Check on my hardwareAdd to compare →Ollama →Hugging Face →

Model size→Hardware requirement→Actual fit

Granite 4.2 8B needs ~6.5–11.3 GB of VRAM depending on quantization. The bar is what it needs; the marker is the card's VRAM. Speed figures below are planning estimates.

8 GB

RTX 4060 · 3060 Ti

6.5
8
Q4_K_M · Est. ~23 tok/sComfortable

12 GB

RTX 4070 · 3060 12G

7.5
12
Q5_K_M · Est. ~34 tok/sComfortable

16 GB

RTX 4080 · M-series 16G

11.3
16
Q8_0 · Est. ~35 tok/sComfortable

24 GB

RTX 4090 · 3090

11.3
24
Q8_0 · Est. ~53 tok/sComfortable

48 GB

A6000 · 2×24G

11.3
48
Q8_0 · Est. ~53 tok/sComfortable

02  — Quantizations

QuantFile sizeEst. VRAM
Q4_K_M5.3 GB6.5 GB
Q5_K_M6.3 GB7.5 GB
Q8_010.1 GB11.3 GB

03  — Runtimes & use

llama.cpp · Ollama · LM Studio

General · Fast

Memory and speed figures are planning estimates, not measured benchmarks. Memory includes a fixed allowance for KV cache and runtime overhead. Long-context prompts push VRAM use above the numbers shown here.