Llama 3.1 8B VRAM requirements

Can I run Llama 3.1 8B on…

• RTX 4090: Yes, up to BF16
• RTX 3090: Yes, up to BF16
• RTX 5090: Yes, up to BF16
• RTX 3060 12GB: Yes, up to Q5_K_M
• RTX 4060 Ti 16GB: Yes, up to Q8_0
• RTX 5060 Ti 16GB: Yes, up to Q8_0
• RTX 4070 12GB: Yes, up to Q5_K_M
• RTX 5080 16GB: Yes, up to Q8_0
• RTX 4080 Super 16GB: Yes, up to Q8_0
• RX 7900 XTX: Yes, up to BF16

How much VRAM does Llama 3.1 8B need?

Llama 3.1 8B needs 6.58 GB at Q4_K_M, 10.3 GB at Q8_0 and 18.1 GB at FP16 with 8,192 tokens of context and one request. Each total below is the weights plus the FP16 KV cache and the runtime overhead (0.5 GB plus 10%), and each row opens the calculator with that setting. What the GGUF names mean.

Fine-tuning Llama 3.1 8B? Llama 3.1 8B VRAM for LoRA, QLoRA and full training.

Llama 3.1 8B GGUF files on Hugging Face

The 7 GGUF files llama.cpp picks for Llama 3.1 8B with -hf <repo>:<QUANT>, with the bytes the Hugging Face file list reported on 2026-09-29. Each total is that file plus the FP16 KV cache and 0.5 GB + 10%, one request. "vs estimate" compares the file with the bits-per-weight size the table above uses.

GB here is GiB (1024³ bytes), as everywhere on this site; Hugging Face shows the Q4_K_M file as 4.9 GB, in 10⁹ bytes.

Which Llama 3.1 8B GGUF fits a 16, 24, 32, 48 or 80 GB GPU or a Mac

From the file sizes above: the largest file that fits each machine with an 8K context and 0.5 GB left free, the longest context it then has room for, and the longest context for Q4_K_M (4.58 GB). One request, FP16 KV cache.

Run Llama 3.1 8B with llama.cpp or Ollama

llama-server -hf bartowski/Meta-Llama-3.1-8B-Instruct-GGUF:Q4_K_M -c 131072

Meta-Llama-3.1-8B-Instruct-Q4_K_M.gguf, 4.9 GB, from bartowski/Meta-Llama-3.1-8B-Instruct-GGUF (checked 2026-09-29). On a 24 GB RTX 3090 / 4090, Q4_K_M with -c 131072 takes 23.1 GB and leaves 880 MB.

ollama run llama3.1:8b

Ollama library tag llama3.1:8b; its quant is Ollama's choice. Same context: /set parameter num_ctx 131072.

Llama 3.1 8B at 8K, 32K and 128K (full) tokens of context

All 32 layers use full attention. Each extra token of context adds 128 KB of FP16 cache per request. The last column is the smallest setup here, one card or a group, that holds it at Q4_K_M. How the KV cache works.

How fast Llama 3.1 8B writes

Tokens per second for one request with 8,192 tokens of context, estimated from memory bandwidth. A dash means it does not fit on one card. Try other settings in the speed calculator.

Longest context on one GPU

How many tokens of context Llama 3.1 8B fits on a single card with one request, an FP16 KV cache and 0.5 GB left free. "Full" means the model's whole context window fits.

Model details

Parameters

8.0B (8,030,261,248)

Layers

All 32 layers use full attention

Attention cache

8 KV heads × 128

Context length

131,072 tokens

Published weights

15.0 GB (BF16)

On Hugging Face

meta-llama/Llama-3.1-8B-Instruct

Reported benchmarks

GPQA, 0-shot 30.4 · MMLU-Pro (CoT), 5-shot 48.3. Vendor-reported: from the maker’s own model card, not rerun here; read 2026-09-30.

Llama 3.1 8B next to similar-sized models

The three models here closest to it in parameter count, at Q4_K_M with 8,192 tokens of context.