Skip to content
localmodel.run

Text model · llama

Llama 3.1 8B requirements

llama family · 8B params · released Jul 2024 · 111M Ollama pulls · LMArena Elo 1211. Minimum to run at Q4_K_M: Nvidia GeForce RTX 3060 (12GB).

LicenseLlama 3.1 Community· Conditional↓ 6.6M/mo♥ 6.1Kon HuggingFace
Q4_K_M
4.92 GB
Q8_0
8.54 GB
Total @ Q4 (4k)
~6.4 GB
Context
128 k

Quantization sizes

GGUF quantson disk
QuantizationSize on disk
Q2_K3.4 GB est
Q3_K_M3.9 GB est
Q4_K_M (default)4.92 GB
Q5_K_M5.7 GB est
Q6_K6.6 GB est
Q8_08.54 GB
FP1616 GB est

Lower quant = smaller and faster, slightly lower quality. Q4_K_M is the common default.

Run it

Ollama
$ ollama run llama3.1:8b
llama.cpp
$ llama-cli -hf bartowski/Meta-Llama-3.1-8B-Instruct-GGUF:Q4_K_M
LM Studio
$ lms get bartowski/Meta-Llama-3.1-8B-Instruct-GGUF

Which devices can run Llama 3.1 8B?

FAQ

How much VRAM or RAM does Llama 3.1 8B need?

At Q4_K_M, Llama 3.1 8B needs about 6.4 GB (weights ~4.92 GB + KV cache + overhead) at a 4k context. At Q8_0 budget ~10 GB.

Can Llama 3.1 8B run on a laptop?

Yes, Llama 3.1 8B fits on a 16 GB machine at Q4_K_M and runs on Apple Silicon or a 12 GB+ GPU comfortably.

Can I use Llama 3.1 8B commercially?

Conditionally. Llama 3.1 Community License: free under 700M MAU.

Q4_K_M and Q8_0 sizes from bartowski HF repo. Default ollama tag pulls 4.9GB Q4_K_M. Min RAM estimate includes model + small KV cache overhead.

Sources

Memory figures are estimates. See methodology.