MiniMax H3 (33B, video + audio) on AMD Instinct MI350X

Can the AMD Instinct MI350X run MiniMax H3 (33B, video + audio)?

Yes, but slowly

Slow. At FP16 it needs about 113 GB of 288 GB usable and takes about 17.8 min per clip.

Best precision: FP16
Memory mode: Everything on GPU
Time per clip: 17.8 min
Output: 1344×768, 124 frames, 50 steps

Best app to run it on the AMD Instinct MI350X

On Linux

ComfyUI

Desktop app · runs on PyTorch (ComfyUI), GGUF via ComfyUI-GGUF

Runs this model with built-in or popular custom nodes; GGUF files load with ComfyUI-GGUF.

Get ComfyUI

On Windows

ComfyUI

Desktop app · runs on PyTorch (ComfyUI), GGUF via ComfyUI-GGUF

Runs this model with built-in or popular custom nodes; GGUF files load with ComfyUI-GGUF.

Get ComfyUI

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 MiniMax H3 (33B, video + audio)

Quantized files · GGUF for ComfyUI

Smaller GGUF versions for ComfyUI (with the ComfyUI-GGUF node). No sign-in needed; the model’s license still applies.

huggingface.co/unsloth/MiniMax-H3-GGUF (opens in a new tab)

Original weights · Hugging Face

Full-precision files from MiniMax, for fine-tuning, vLLM, Transformers, Diffusers or ComfyUI.

huggingface.co/MiniMaxAI/MiniMax-H3 (opens in a new tab)

No sign-in needed.

Official repo · GitHub

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

github.com/MiniMax-AI/MiniMax-H3 (opens in a new tab)

git clone https://github.com/MiniMax-AI/MiniMax-H3

Links checked 3 days ago against Hugging Face, Ollama and GitHub.

How it’s rated

Quality 90Excellent quality (A+) · 90 of 100: how capable the model is at FP16, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking.: How capable the model itself is at FP16, 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 Slow: 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.	40.5 / 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.	6.2 / 15
Memory useRewards using the memory you have, so a bigger, better variant beats a tiny one that leaves most memory idle.	9.0 / 15
Match score	80.7 / 100

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

Every precision on the AMD Instinct MI350X

Quant	Memory needed	Available	Speed	Memory mode	Runs	Quality	Download
`FP16`	113 GB	288 GB	17.8 min	Everything on GPU	Slow	90Excellent quality (A+) · 90 of 100: how capable the model is at FP16, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking.	Original weights (opens in a new tab)
`FP8`	66.8 GB	288 GB	16.0 min	Everything on GPU	Slow	88Excellent quality (A+) · 88 of 100: how capable the model is at FP8, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.	Original weights (opens in a new tab)
`Q8 GGUF`	68.2 GB	288 GB	20.5 min	Everything on GPU	Slow	89Excellent quality (A+) · 89 of 100: how capable the model is at Q8 GGUF, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q4 GGUF`	58.7 GB	288 GB	21.3 min	Everything on GPU	Slow	83Great quality (A) · 83 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)

Other hardware that runs it

88Excellent quality (A+) · 88 of 100: how capable the model is at FP8, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA RTX PRO 6000 Blackwell FP8 · 11.4 min · Runs well
90Excellent quality (A+) · 90 of 100: how capable the model is at FP16, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA B200 FP16 · 10.6 min · Runs well
88Excellent quality (A+) · 88 of 100: how capable the model is at FP8, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GH200 Grace Hopper (96 GB GPU) FP8 · 10.9 min · Runs well
88Excellent quality (A+) · 88 of 100: how capable the model is at FP8, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA H100 80GB SXM FP8 · 10.9 min · Runs well
88Excellent quality (A+) · 88 of 100: how capable the model is at FP8, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA H800 80GB SXM FP8 · 10.9 min · Runs well
88Excellent quality (A+) · 88 of 100: how capable the model is at FP8, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA H100 NVL FP8 · 11.8 min · Runs well

Quick answers

How much memory does MiniMax H3 (33B, video + audio) need on the AMD Instinct MI350X 288GB?

At FP16 it needs about 113 GB. The AMD Instinct MI350X 288GB has 288 GB usable.

How long does MiniMax H3 (33B, video + audio) take on the AMD Instinct MI350X 288GB?

About 17.8 min per clip at FP16 and default settings, an estimate scaled from a reference timing by this hardware’s compute speed.

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.