Cosmos3-Super Text2Image (64B) on Orange Pi AIpro (20T) (12 GB)

Can the Orange Pi AIpro (20T) (12 GB) run Cosmos3-Super Text2Image (64B)?

No, it doesn’t fit

No. Cosmos3-Super Text2Image (64B) needs about 26.1 GB even at FP8; the Orange Pi AIpro (20T) (12 GB) has 9.8 GB usable.

Download Cosmos3-Super Text2Image (64B)

Original weights · Hugging Face

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

huggingface.co/nvidia/Cosmos3-Super-Text2Image (opens in a new tab)

No sign-in needed.

Official repo · GitHub

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

github.com/nvidia/cosmos (opens in a new tab)

git clone https://github.com/nvidia/cosmos

Links checked 7 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 FP8, whatever it runs on. Match score 12/100 (quality, fit, speed and memory use) sets the ranking.: How capable the model itself is at FP8, 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 Won't run: How well it runs on the Orange Pi AIpro (20T) (12 GB): 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.	12.2 / 45
Memory fitFull marks when the model fits in the share of unified memory the GPU can use; fewer when it goes past that limit.	0.0 / 25
SpeedGeneration speed against a 40 tokens/s target for LLMs, or render time against a target for image and video.	0.0 / 15
Memory useRewards using the memory you have, so a bigger, better variant beats a tiny one that leaves most memory idle.	0.0 / 15
Match score	12.2 / 100

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

It doesn’t fit, so only a reduced share of the quality points counts and fit, speed and memory use score zero.

Every precision on the Orange Pi AIpro (20T) (12 GB)

Quant	Memory needed	Available	Speed	Memory mode	Runs	Quality	Download
`FP16`	48.5 GB	9.8 GB	—	—	Won't run	92Excellent quality (A+) · 92 of 100: how capable the model is at FP16, whatever it runs on. Match score 12/100 (quality, fit, speed and memory use) sets the ranking.	Original weights (opens in a new tab)
`FP8`	26.1 GB	9.8 GB	—	—	Won't run	90Excellent quality (A+) · 90 of 100: how capable the model is at FP8, whatever it runs on. Match score 12/100 (quality, fit, speed and memory use) sets the ranking.	Original weights (opens in a new tab)

Hardware that can run it

90Excellent quality (A+) · 90 of 100: how capable the model is at FP8, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA RTX PRO 6000 Blackwell FP8 · 26 s · Runs well
90Excellent quality (A+) · 90 of 100: how capable the model is at FP8, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 5090 FP8 · 39 s · Runs well (CPU offload)
92Excellent quality (A+) · 92 of 100: how capable the model is at FP16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA B200 FP16 · 24 s · Runs well
90Excellent quality (A+) · 90 of 100: how capable the model is at FP8, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GH200 Grace Hopper (96 GB GPU) FP8 · 25 s · Runs well
90Excellent quality (A+) · 90 of 100: how capable the model is at FP8, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA H100 NVL FP8 · 27 s · Runs well
90Excellent quality (A+) · 90 of 100: how capable the model is at FP8, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA RTX PRO 6000 Blackwell Server Edition FP8 · 27 s · Runs well

Models that do run on the Orange Pi AIpro (20T) (12 GB)

90Excellent quality (A+) · 90 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. FLUX 3 Image Klein Q8 GGUF · 28.3 min
72Great quality (A) · 72 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. FLUX.2 [klein] 4B Q4 GGUF · 4.9 min
81Great quality (A) · 81 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. ERNIE-Image Q4 GGUF · 1.9 h
81Great quality (A) · 81 of 100: how capable the model is at Q8 GGUF, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. MOSS-Image SFT Q8 GGUF · 69.5 min
83Great quality (A) · 83 of 100: how capable the model is at Q8 GGUF, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. Z-Image Turbo Q8 GGUF · 12.4 min

Quick answers

How much memory does Cosmos3-Super Text2Image (64B) need on the Orange Pi AIpro (20T) 12GB?

Even at FP8 it needs about 26.1 GB; the Orange Pi AIpro (20T) 12GB has 9.8 GB usable.

What can the Orange Pi AIpro (20T) 12GB run instead?

FLUX 3 Image Klein, FLUX.2 [klein] 4B and ERNIE-Image run best on it; the page lists more, each with its speed.

What hardware can run Cosmos3-Super Text2Image (64B)?

NVIDIA RTX PRO 6000 Blackwell 96GB, RTX 5090 32GB and NVIDIA B200 180GB run it; the page lists more, each with its 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.