Rent a GPU

Granite 3.3 8B hardware requirements

8.17B parameters. Needs about 6.9 GB at Q4_K_M with 8K context.

Parameters8.17BDense
Q4_K_M · 8K6.9 GBthe usual download
FP16 · 8K18.3 GBfull precision
Runs well on51/51GPUs and Macs we track

Memory needed at each quantization

Weights, KV cache and runtime overhead, at 8K and 32K tokens of context.

QuantQuality8K context32K context
FP16Full quality18.3 GB22.3 GB
Q8_0Near-lossless10.6 GB14.6 GB
Q6_KExcellent8.6 GB12.7 GB
Q5_K_MVery good7.8 GB11.8 GB
Q4_K_MGood, most popular6.9 GB10.9 GB
Q3_K_MNoticeably degraded5.9 GB10 GB
Q2_KHeavily degraded5 GB9 GB

Run it with Ollama

Downloads Q4_K_M by default.

ollama run granite3.3:8b

Model card

ibm-granite/granite-3.3-8b-instruct on Hugging Face: licence, prompt format and GGUF builds.

Run it in the cloud

Run Granite 3.3 8B on a rented GPU today.

It needs about 6.9 GB at Q4_K_M: a rented RTX 3090 / RTX 4090 runs it for about $0.18 to $0.89 an hour.

RunPodRent a GPU on RunPodVast.aiRent a GPU on Vast.ai

Every GPU and Mac checked

Sorted by memory. Bars show how much of each card Granite 3.3 8B fills at the best quant that fits.

51 Runs well

NameMemory usedBest quantSpeed