Llama 3.1 405B Instruct on AMD Instinct MI350X

Can the AMD Instinct MI350X run Llama 3.1 405B Instruct?

Yes, it runs well (tight fit)

Runs well (tight fit). At Q4_K_M it needs about 267 GB of 288 GB usable, with an estimated speed of 20.8 tok/s.

Best quant: Q4_K_M
Generation speed: 20.8 tok/s
Prompt processing: 209 tok/s
First token (1K prompt): 4.9 s
Largest safe context: 48K

Best app to run it on the AMD Instinct MI350X

On Linux

Lemonade

Desktop app · runs on llama.cpp, ONNX Runtime GenAI (Ryzen AI NPU)

AMD’s own app: llama.cpp tuned for Radeon on Vulkan and ROCm.

lemonade pull MaziyarPanahi/Meta-Llama-3.1-405B-Instruct-GGUF:Q4_K_M

LM Studio Point-and-click: search, download and chat in one window.
Ollama One command with Ollama’s own build of this model.

On Windows

Lemonade

Desktop app · runs on llama.cpp, ONNX Runtime GenAI (Ryzen AI NPU)

AMD’s own app: llama.cpp tuned for Radeon on Vulkan and ROCm.

lemonade pull MaziyarPanahi/Meta-Llama-3.1-405B-Instruct-GGUF:Q4_K_M

LM Studio Point-and-click: search, download and chat in one window.
Ollama One command with Ollama’s own build of this model.

How we choose: the files published for this model, your operating system and GPU, whether it fits fully in memory, and published speed tests (MLX vs llama.cpp on Apple silicon, vLLM vs llama.cpp vs Ollama on NVIDIA). Details.

Download Llama 3.1 405B Instruct

Recommended for the AMD Instinct MI350X: Q4_K_M. Every other quantization has its own download link in the table below.

Quickest · Ollama

One command downloads a ready-to-run version and starts a chat.

Get Ollama at ollama.com/download (opens in a new tab) · model page (opens in a new tab)

ollama run llama3.1:405b-instruct-q4_K_M

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

huggingface.co/MaziyarPanahi/Meta-Llama-3.1-405B-Instruct-GGUF (opens in a new tab)

ollama run hf.co/MaziyarPanahi/Meta-Llama-3.1-405B-Instruct-GGUF

Original weights · Hugging Face

Full-precision files from Meta, for fine-tuning, vLLM, Transformers.

huggingface.co/meta-llama/Llama-3.1-405B-Instruct (opens in a new tab)

Free Hugging Face account needed: request access on the model page, and Meta approves it (usually within minutes to a few hours).

Official repo · GitHub

Reference code, inference scripts and docs from Meta. The weights themselves download from Hugging Face.

github.com/meta-llama/llama-models (opens in a new tab)

Meta’s official Llama repo: model cards, prompt formats and download tools.

git clone https://github.com/meta-llama/llama-models

Links checked 1 day ago against Hugging Face, Ollama and GitHub.

How it’s rated

Quality 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking.: How capable the model itself is at Q4_K_M, from its editorial quality tier scaled by the quantization or precision. It’s the same on any hardware that runs this variant. Scale: A+ 85+ · A 70+ · B 55+ · C 40+ · D under 40.
Runs Runs welltight fit: How well it runs on the AMD Instinct MI350X: speed, capped by how well it fits in memory. Runs great: 30+ tokens/s (or near the image, video or decision speed target) · Runs well: 10+ tokens/s · Slow: 3+ tokens/s · Very slow: under 3 · Won’t run: doesn’t fit even with offload. A tight fit or RAM offload caps it at Runs well; heavy offload caps it at Slow.

Match score

To order lists, each result also gets a match score out of 100 from four parts. It favors the most capable model that still runs well.

Model qualityEditorial quality tier of the model, scaled by the quantization or precision used.	30.8 / 45
Memory fitFull marks when everything sits in GPU memory; fewer when part spills to system RAM.	24.0 / 25
SpeedGeneration speed against a 40 tokens/s target for LLMs, or render time against a target for image and video.	10.8 / 15
Memory useRewards using the memory you have, so a bigger, better variant beats a tiny one that leaves most memory idle.	15.0 / 15
Match score	80.6 / 100

The match score orders results; it is not shown as a grade. Quality and Runs are the two ratings in the tables.

Every quantization on the AMD Instinct MI350X

Quant	Memory needed	Available	Speed	Runs	Quality	Download
`Q2_K`	182 GB	288 GB	34.2 tok/s	Runs great	50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.	GGUF · part 1 of 9 (opens in a new tab)
`Q3_K_M`	218 GB	288 GB	26.9 tok/s	Runs well	60Good quality (B) · 60 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (Q3_K_S) · part 1 of 9 (opens in a new tab)
`Q4_K_M`	267 GB	288 GB	20.8 tok/s	Runs welltight fit	68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking.
`Q5_K_M`	307 GB	288 GB	2.2 tok/s	Very slowoffloads to RAM	69Good quality (B) · 69 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking.
`Q6_K`	352 GB	288 GB	—	Won't run	71Great quality (A) · 71 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking.
`Q8_0`	449 GB	288 GB	—	Won't run	72Great quality (A) · 72 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking.
`F16`	831 GB	288 GB	—	Won't run	72Great quality (A) · 72 of 100: how capable the model is at F16, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking.	Original weights (opens in a new tab)

Memory figures assume 8K of context. Longer conversations or documents grow the KV cache.

Other hardware that runs it

60Good quality (B) · 60 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip Q3_K_M · 27.8 tok/s · Runs well (tight fit)
60Good quality (B) · 60 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. AMD Instinct MI325X Q3_K_M · 20.2 tok/s · Runs well
50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA B200 Q2_K · 39.7 tok/s · Runs well (tight fit)
50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. AMD Instinct MI300X Q2_K · 22.7 tok/s · Runs well (tight fit)
50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. Apple M3 Ultra (256 GB) Q2_K · 3.5 tok/s · Slow (tight fit)
50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. Apple M3 Ultra (512 GB) Q2_K · 3.5 tok/s · Slow

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.