Llama 4 Maverick (17B x 128E) on AMD Instinct MI325X

Can the AMD Instinct MI325X run Llama 4 Maverick (17B x 128E)?

Yes, it runs great

Runs great. At Q3_K_M it needs about 212 GB of 256 GB usable, with an estimated speed of 188 tok/s.

Best quant: Q3_K_M
Generation speed: 188 tok/s
Prompt processing: 5,959 tok/s
First token (1K prompt): 0.3 s
Largest safe context: 227K

Best app to run it on the AMD Instinct MI325X

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 unsloth/Llama-4-Maverick-17B-128E-Instruct-GGUF:Q3_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 unsloth/Llama-4-Maverick-17B-128E-Instruct-GGUF:Q3_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.

Want an agent? Llama 4 Maverick (17B x 128E) supports tool use, so you can pair a llama.cpp or MLX LM server with one of these. Apps that run on this computer come first.

Coding agents

Personal assistants

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 4 Maverick (17B x 128E)

Recommended for the AMD Instinct MI325X: Q3_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 llama4:maverick

Quantized files · GGUF

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

Download the Q3_K_M files (opens in a new tab)

huggingface.co/unsloth/Llama-4-Maverick-17B-128E-Instruct-GGUF (opens in a new tab)

ollama run hf.co/unsloth/Llama-4-Maverick-17B-128E-Instruct-GGUF

Original weights · Hugging Face

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

huggingface.co/meta-llama/Llama-4-Maverick-17B-128E-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 60Good quality (B) · 60 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking.: How capable the model itself is at Q3_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 great: How well it runs on the AMD Instinct MI325X: 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.	27.2 / 45
Memory fitFull marks when everything sits in GPU memory; fewer when part spills to system RAM.	25.0 / 25
SpeedGeneration speed against a 40 tokens/s target for LLMs, or render time against a target for image and video.	15.0 / 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	82.2 / 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 MI325X

Quant	Memory needed	Available	Speed	Runs	Quality	Download
`Q2_K`	176 GB	256 GB	210 tok/s	Runs great	50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking.	GGUF · folder (opens in a new tab)
`Q3_K_M`	212 GB	256 GB	188 tok/s	Runs great	60Good quality (B) · 60 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking.	GGUF · folder (opens in a new tab)
`Q4_K_M`	260 GB	256 GB	107 tok/s	Runs welloffloads to RAM	68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking.	GGUF · folder (opens in a new tab)
`Q5_K_M`	300 GB	256 GB	—	Won't run	69Good quality (B) · 69 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking.	GGUF · folder (opens in a new tab)
`Q6_K`	344 GB	256 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.	GGUF · folder (opens in a new tab)
`Q8_0`	441 GB	256 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.	GGUF · folder (opens in a new tab)
`F16`	818 GB	256 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.	GGUF · folder (opens in a new tab) 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

69Good quality (B) · 69 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. Apple M3 Ultra (512 GB) Q5_K_M · 36.7 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. Apple M3 Ultra (256 GB) Q2_K · 66.3 tok/s · Runs well (tight fit)
69Good quality (B) · 69 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. Apple M5 Ultra (512 GB) Q5_K_M · 50.7 tok/s · Runs great
68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. AMD Instinct MI350X Q4_K_M · 186 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 82/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip Q3_K_M · 211 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. AMD Instinct MI300X Q2_K · 201 tok/s · Runs well (tight fit)

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.