Nous ResearchLlama 3LlamaForCausalLM

Meta Llama 3.1 8B Instruct — Hardware Requirements & GPU Compatibility

Chat

Meta Llama 3.1 8B Instruct is a 8.0B-parameter open language model from Nous Research in the Llama 3 family. It supports a context window of up to 131,072 tokens. At Q4_K_M it needs about 5.39 GB of VRAM — see which GPUs and Macs can run it below.

8.0B params · dense131K contextLlama 3.1 CommunityJul 2024186.5K downloads40 likes893 quant downloads

Your hardware

Verdicts below are for the reference RTX 4090. Set yours to personalise the page.

Detection reads your browser's GPU name with no data sent anywhere.

Runs great on a 24 GB GPU

Runs from 8 GB (Q4_K_M · 16K)Best on 24 GB (Q4_K_M · ~122 tok/s · 128K)Macs from 32 GB (~16 tok/s)

Llama 3.1 Community

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Meta Llama 3.1 8B Instruct

One recommended way. Every other method, OS and quant is in the step-by-step guide.

Other ways

each opens step-by-step

Run in cloud

Fits on RTX 3060 12GB (4 GB headroom) · Q4_K_M

Generation speed: ~43 tok/s; generation speed
Cost per 1M output tokens: $0.44; per 1M output tokens

Deployon Vast.aiBest optionfrom $0.069/hr

Compare GPUs →

Run Meta Llama 3.1 8B Instruct

7.3 GB VRAM at 16K context · fits the RTX 4090 (16.7 GB free)

Works on macOS, Windows and Linux.

Install Ollama (Linux)
$curl -fsSL https://ollama.com/install.sh | sh
Install Ollama (macOS / Windows)
Download and run the app; it installs the ollama command.
Download Ollama ↗
Run Q4_K_M
$ollama run hf.co/featherless-ai-quants/NousResearch-Meta-Llama-3.1-8B-Instruct-GGUF:Q4_K_M
Raise the context window
Type it inside the chat. For the API server instead, start it with OLLAMA_CONTEXT_LENGTH=16384 ollama serve.
$/set parameter num_ctx 16384
A larger context needs more memory than the weights alone.
Use the API
Ollama serves its API on http://localhost:11434.

Works on macOS, Windows and Linux.

Install LM Studio
Desktop app for macOS (Apple silicon), Windows and Linux.
lmstudio.ai ↗
Download the model
In the app press Ctrl/Cmd + Shift + M and paste https://huggingface.co/featherless-ai-quants/NousResearch-Meta-Llama-3.1-8B-Instruct-GGUF, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get https://huggingface.co/featherless-ai-quants/NousResearch-Meta-Llama-3.1-8B-Instruct-GGUF@Q4_K_M
Open on Hugging Face ↗
Load and chat
Open the model loader, select the model and set Context Length to 16384. Leave GPU offload at maximum if it fits.
Start the local server (optional)
$lms server start

Works on macOS, Windows and Linux.

Install llama.cpp (macOS)
MacPorts, Nix and conda-forge also work.
$brew install llama.cpp
Install options ↗
Install llama.cpp (Windows)
$winget install llama.cpp
Install llama.cpp (Linux)
Download the prebuilt llama-*-bin-ubuntu-* archive for your GPU backend (CPU, Vulkan) from the releases page, unpack it and run the binaries from that folder. Homebrew (brew install llama.cpp) and conda-forge also work on Linux.
llama.cpp releases ↗
Start the server
Downloads the model on first run. The chat UI and an OpenAI-compatible API are at http://localhost:8080.
$llama-server -hf featherless-ai-quants/NousResearch-Meta-Llama-3.1-8B-Instruct-GGUF:Q4_K_M -c 16384 -ngl 99 --port 8080

Works on macOS, Windows and Linux.

Option A: Docker Model Runner
Docker Desktop: Settings > AI > Enable Docker Model Runner. Docker Engine on Linux: install the docker-model-plugin package.
$docker model run hf.co/featherless-ai-quants/NousResearch-Meta-Llama-3.1-8B-Instruct-GGUF:Q4_K_M
Model Runner docs ↗
The quant tag after the repo is how Hugging Face cards write it; if your Docker version rejects it, drop the tag and pick the quant by repo.
Option B: llama.cpp server image (NVIDIA)
Needs the NVIDIA Container Toolkit. Models are cached in the llama-cache volume. Open http://localhost:8080.
$docker run --gpus all -p 8080:8080 -v llama-cache:/root/.cache ghcr.io/ggml-org/llama.cpp:server-cuda -hf featherless-ai-quants/NousResearch-Meta-Llama-3.1-8B-Instruct-GGUF:Q4_K_M -c 16384 -ngl 99 --host 0.0.0.0 --port 8080
Image tags (cuda13, vulkan, rocm) ↗
Cache path inside the image is not in the docs; remove the -v flag if the volume complains.

Linux only.

Install vLLM
Linux only; on Windows run it inside WSL2. NVIDIA GPU shown; AMD ROCm, Intel and CPU builds have their own install pages.
$uv venv --python 3.12 --seed
$source .venv/bin/activate
$uv pip install vllm --torch-backend=auto
vLLM installation ↗
Serve the model
OpenAI-compatible API at http://localhost:8000/v1.
$vllm serve NousResearch/Meta-Llama-3.1-8B-Instruct --max-model-len 16384
Runs the original ~18 GB weights, not the Q4_K_M GGUF.

macOS only.

MLX supports only macOS, so follow these steps for that setup.

Install mlx-lm
$pip install mlx-lm
Chat in the terminal
$mlx_lm.chat --model ChetanT271/Llama-3.1-8B-Instruct-MLX-4bit
Downloads the model on first run. The MLX repo has its own quantisation; it may not match the quant selected above.
Or start a local server
Listens on http://localhost:8080.
$mlx_lm.server --model ChetanT271/Llama-3.1-8B-Instruct-MLX-4bit

Works on macOS, Windows and Linux.

Install Jan
jan.ai ↗
Find the model
Open the Hub tab, paste this model ID in Search, and choose Q4_K_M if offered.
$featherless-ai-quants/NousResearch-Meta-Llama-3.1-8B-Instruct-GGUF
Use the local API (optional)
Settings > Local API Server > Start Server. It listens on http://127.0.0.1:1337.

Rent GPUs hourly to run Q4_K_M right there. Pricing comes from partner marketplaces and varies.

Available GPU rentals for Q4_K_M, prioritized by value
GPU	Speed	$/1M tokens	Rent
RTX 3060 12GBCheapestCheapest4 GB free	~43 tok/s	$0.44	Deploy on Vast.aifrom $0.069/hr (opens in new tab)	4 GB free · ~43 tok/s · $0.44 / 1M tokens
RTX A4000 (16 GB)8 GB free	~54 tok/s	$0.58	Deploy on Vast.aifrom $0.11/hr (opens in new tab)	16 GB · 8 GB free · ~54 tok/s · $0.58 / 1M tokens
RTX 4060 Ti 16GB8 GB free	~35 tok/s	$1.28	Deploy on Vast.aifrom $0.16/hr (opens in new tab)	8 GB free · ~35 tok/s · $1.28 / 1M tokens
RTX 2080 Ti (11 GB)3 GB free	~74 tok/s	$0.63	Deploy on Vast.aifrom $0.17/hr (opens in new tab)	11 GB · 3 GB free · ~74 tok/s · $0.63 / 1M tokens
RTX 4000 Ada (20 GB)12 GB free	~43 tok/s	$1.28	Deploy on RunPodfrom $0.20/hr (opens in new tab)	20 GB · 12 GB free · ~43 tok/s · $1.28 / 1M tokens
RTX 3090 (24 GB)16 GB free	~113 tok/s	$0.54	Deploy on RunPodfrom $0.22/hr (opens in new tab)	24 GB · 16 GB free · ~113 tok/s · $0.54 / 1M tokens
RTX PRO 4500 (32 GB)24 GB free	~108 tok/s	$0.87	Deploy on RunPodfrom $0.34/hr (opens in new tab)	32 GB · 24 GB free · ~108 tok/s · $0.87 / 1M tokens
RTX 4090 (24 GB)16 GB free	~122 tok/s	$0.78	Deploy on RunPodfrom $0.34/hr (opens in new tab)	24 GB · 16 GB free · ~122 tok/s · $0.78 / 1M tokens

Compare GPUs →

How Much VRAM Does Meta Llama 3.1 8B Instruct Need?

Select a quantization level; the listed hardware adheres to it.

VRAM	+32K	+128K	File	Quality	On RTX 4090 (reference)
3.3 GB	7.3 GB	20.2 GB	2.9 GB	2-bit · i-quant medium — Importance-weighted 2-bit, medium	Fits: 128K context · ~200 tok/s
4.0 GB	8 GB	20.9 GB	3.2 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Fits: 128K context · ~165 tok/s
4.1 GB	8.1 GB	21 GB	3.7 GB	3-bit small — 3-bit small quantization	Fits: 128K context · ~161 tok/s
4.2 GB	8.2 GB	21.1 GB	3.8 GB	3-bit · i-quant medium — Importance-weighted 3-bit, medium	Fits: 128K context · ~157 tok/s
4.5 GB	8.5 GB	21.4 GB	4 GB	3-bit medium — 3-bit medium quantization	Fits: 128K context · ~146 tok/s
4.7 GB	8.7 GB	21.6 GB	4.3 GB	3-bit large — 3-bit large quantization	Fits: 128K context · ~140 tok/s
4.9 GB	8.9 GB	21.8 GB	4.5 GB	4-bit · i-quant compact — Importance-weighted 4-bit, compact	Fits: 128K context · ~134 tok/s
5.1 GB	9.1 GB	22 GB	4.7 GB	4-bit small — 4-bit small quantization	Fits: 128K context · ~129 tok/s
5.4 GB	9.4 GB	22.3 GB	4.9 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Fits: 128K context · ~122 tok/s
5.6 GB	9.6 GB	22.5 GB	5.6 GB	5-bit · legacy — 5-bit legacy quantization	Fits: 128K context · ~117 tok/s
6.1 GB	10.1 GB	23 GB	5.6 GB	5-bit small — 5-bit small quantization	Fits: 128K context · ~108 tok/s
6.3 GB	10.3 GB	23.2 GB	5.7 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Fits: 128K context · ~104 tok/s
7.2 GB	11.2 GB	24.1 GB	6.6 GB	6-bit, very good — 6-bit quantization, very good quality	Fits: 64K context · ~91 tok/s
8.6 GB	12.6 GB	25.5 GB	8.5 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Fits: 64K context · ~76 tok/s
16.6 GB	20.7 GB	33.5 GB	~16.1 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Fits: 32K context · ~39 tok/s

est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

7.3GBat 16K context

Weights: 4.8 GB
KV cache: 2.2 GB
Overhead: 0.3 GB

Fits RTX 4090 (reference) · 16.7 GB free · ~122 tok/s

≥50 tok/s needs ≥414 GB/s · RTX 5090 ~216, RTX 3090 Ti ~122

Which Hardware Can Run Meta Llama 3.1 8B Instruct?

Q4_K_M · 5.4 GB

Meta Llama 3.1 8B Instruct (Q4_K_M) requires 5.4 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 8+ GB is recommended. Using the full 131K context window can add up to 16.9 GB, bringing total usage to 22.3 GB. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti, NVIDIA GeForce RTX 3070 Ti. 159 devices with unified memory can run Meta Llama 3.1 8B Instruct, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, NVIDIA Jetson Orin Nano 8GB (Super).

By VRAM tier · typical GPUs and Macs at Q4_K_M

Typical hardwareQuantVRAMtok/sContext that fits

8 GB · RTX 4060Q4_K_M · 7.3 GB · ~33 tok/s · 16K7.3~3316K8 GB tier →

12 GB · RTX 3060 12GBQ4_K_M · 7.3 GB · ~43 tok/s · 32K7.3~4332K12 GB tier →

16 GB · RTX 5060 Ti 16GBQ4_K_M · 7.3 GB · ~54 tok/s · 64K7.3~5464K16 GB tier →

24 GB · RTX 4090referenceQ4_K_M · 7.3 GB · ~122 tok/s · 128K7.3~122128K24 GB tier →

Mac 32 GB · Mac Mini M4Q4_K_M · 7.3 GB · ~16 tok/s · 128K7.3~16128KMacs →

Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 7.3 GB · ~71 tok/s · 128K7.3~71128KMacs →

Alternatives to Meta Llama 3.1 8B Instruct

Llama 3 family · 64 models →

Similar models

6tok/s on RTX 4090

Model	Score	VRAM	tok/s	Why it is similar
ALoDLM 8BAmazon	—	7.6 GB	~119	same size
Llm Jp 4.1 8B Thinkingllm-jp	—	7.6 GB	~115	same size
Ling 3.0 TinyInclusion AI	—	5.5 GB	~364	same size
LFM2.5 8B A1BLiquid AI	—	5.6 GB	~335	same size
BitCPM CANN 8BOpenBMB	—	5.6 GB	~127	same size
Gemma 4 E4BGoogle	—	5.6 GB	~123	same size

Specifications

Publisher: Nous Research
Family: Llama 3
Parameters: 8.0B
Architecture: LlamaForCausalLM
Context Length: 131,072 tokens
Vocabulary Size: 128,256
Release Date: 2024-07-24
License: Llama 3.1 Community

Resources

Frequently Asked Questions

How much VRAM does Meta Llama 3.1 8B Instruct need?

Meta Llama 3.1 8B Instruct requires 5.4 GB of VRAM at Q4_K_M, or 16.6 GB at BF16. Full 131K context adds up to 16.9 GB (22.3 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 8.0B × 4.8 bits ÷ 8 = 4.8 GB

KV Cache + Overhead ≈ 0.6 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 17.5 GB (at full 131K context)

VRAM usage by quantization

Q4_K_M

5.4 GB

Q4_K_M + full context

22.3 GB

Learn more about VRAM estimation →

What's the best quantization for Meta Llama 3.1 8B Instruct?

For Meta Llama 3.1 8B Instruct, Q4_K_M (5.4 GB) offers the best balance of quality and VRAM usage. Q5_0 (5.6 GB) provides better quality if you have the VRAM. The smallest option is IQ2_M at 3.3 GB.

VRAM requirement by quantization

IQ2_M

3.3 GB~62%

Q3_K_M

4.5 GB~83%

Q4_K_S

5.1 GB~88%

Q4_K_M ★

5.4 GB~89%

Q5_K_M

6.3 GB~92%

BF16

16.6 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Meta Llama 3.1 8B Instruct on a Mac?

Yes, but only at lower quantizations. The smallest Mac that can run Meta Llama 3.1 8B Instruct is Mac Mini M1 (8 GB) at IQ2_M; 156 of the 156 Macs we list can run it at some quantization. For Q4_K_M (5.4 GB) you need a Mac with more unified memory.

Can I run Meta Llama 3.1 8B Instruct locally?

Yes — Meta Llama 3.1 8B Instruct can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Meta Llama 3.1 8B Instruct?

At Q4_K_M, Meta Llama 3.1 8B Instruct can reach ~891 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~122 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = (bandwidth GB/s ÷ model GB) × efficiency

Example: NVIDIA B200 → 8000 ÷ 5.39 × 0.65 = ~965 tok/s

Estimated speed at Q4_K_M (5.4 GB)

NVIDIA B200

~965 tok/s

NVIDIA GeForce RTX 4090

~122 tok/s

NVIDIA B300

~965 tok/s

AMD Instinct MI350X

~891 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of Meta Llama 3.1 8B Instruct?

At Q4_K_M, the download is about 4.82 GB. The full-precision BF16 version is 16.06 GB. The smallest option (IQ2_M) is 2.71 GB.

Which GPUs can run Meta Llama 3.1 8B Instruct?

52 consumer GPUs can run Meta Llama 3.1 8B Instruct at Q4_K_M. Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT, AMD Radeon RX 7600. 40 GPUs have plenty of headroom for comfortable inference.

Which devices can run Meta Llama 3.1 8B Instruct?

159 devices with unified memory can run Meta Llama 3.1 8B Instruct at Q4_K_M (5.4 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Apple iPhone 17 Pro, Asus ROG Flow Z13 (2025, Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.