How much VRAM does Wan 2.1 T2V 1.3B need?

At BF16 with everything resident, Wan 2.1 T2V 1.3B peaks at about 15.86 GB. Offloading components one at a time brings that to 5.28 GB, and quantising the weights to 4-bit brings it to 6.27 GB.

What is the cheapest GPU that runs Wan 2.1 T2V 1.3B?

The GeForce GTX 1650 at around $100, running it at FP8 with sequential offload, taking about 30 h 25 min per generation.

Wan · Diffusion transformer

Wan 2.1 T2V 1.3B

Wan 2.1 T2V 1.3B is a pipeline of 3 networks totalling 7.227 billion parameters, of which 1.419 billion do the actual generating. At its native 832x480 and 81 frames the denoiser works on 32,760 latent tokens at once, and attention over that sequence is what sets the render time.

Whole pipeline7.227B

Denoiser1.419B

Download26.9 GB

Default steps50

Native length81 frames

Frame rate16 fps

What the pipeline is made of

This is the part that separates a media model from a language model: these are separate networks, and whether they sit in memory together is your choice, not the model's.

Component	Parameters	Published as	On disk	What it does
transformer	1.419B	F32	5.29 GB	Does the generating. Runs once per step, so it dominates both memory and time.
text_encoder	5.681B	F32	21.16 GB	Turns your prompt into conditioning. Runs once, then can leave the GPU entirely.
vae	0.127B	F32	0.47 GB	Converts between pixels and the compressed latent space. Small, but its decode pass is a memory spike.

These figures were measured from the weight file headers.

Memory by precision

Everything resident, at the model's native output size, on a card large enough that precision is the only constraint.

Precision	Peak VRAM	Quality	What it costs you
BF16	15.86 GB	100%	How the weights are published. No loss, and the largest footprint.
FP8	9.25 GB	97%	Halves the denoiser with a small, usually invisible cost. Needs Ada or newer.
GGUF Q8_0	9.66 GB	98%	Works on any card, unlike FP8. Slightly slower than native precision.
GGUF Q5_K_M	7.34 GB	95%	A middle step when Q8 will not fit.
GGUF Q4_K_M	6.63 GB	91%	The usual way a 12B image model gets onto an 8 GB card. Detail softens.
NF4	6.27 GB	89%	Aggressive 4-bit. Fast to load, noticeably looser on fine detail.

Memory by arrangement

The other lever: the same weights at the same precision, moved around differently. On this model the strongest lever is neither: run the text encoder on the processor. It is 79% of the pipeline, and once it is off the GPU the 20% denoiser is all that remains.

Arrangement	Peak VRAM	Time	How it works
Everything resident	15.86 GB	5 min 28 s	All components stay on the GPU. Fastest, and needs the most memory.
Text encoder on CPU	5.28 GB	5 min 44 s	The prompt is encoded once on the processor, so the encoder never touches the GPU at all. Standard practice on video models, whose encoders are often larger than their denoisers.
Component offload	12.98 GB	6 min 7 s	One component on the GPU at a time. The text encoder runs, then makes way for the denoiser. Costs a few seconds per generation.
Sequential offload	3.99 GB	32 min 40 s	Weights stream layer by layer from system RAM. Runs almost anything on almost anything, and is many times slower.

What a bigger output costs

Both memory and time climb faster than the frame count, because attention is quadratic in the length of the latent sequence.

Output	Latent tokens	Peak VRAM	Time
480p, 3 seconds	20,280	15.29 GB	2 min 16 s
480p, 5 seconds	32,760	15.86 GB	5 min 28 s
720p, 3 seconds	46,800	16.5 GB	10 min 44 s
720p, 5 seconds	75,600	17.82 GB	27 min 0 s
720p, 10 seconds	147,600	21.12 GB	1 h 39 min

Measured at BF16 with everything resident on an RTX 4090, so the columns compare with each other rather than with your machine.

Which hardware runs Wan 2.1 T2V 1.3B

118 of 118 consumer devices run it in some arrangement. Each row shows the fastest arrangement that fits on that device.

Device	Memory	Verdict	Arrangement	Peak	Time
RTX PRO 6000 Blackwell NVIDIA	96 GB	Workable	BF16 everything resident	15.86 GB	3 min 36 s
GeForce RTX 5090 NVIDIA	32 GB	Workable	BF16 everything resident	15.86 GB	4 min 19 s
RTX 6000 Ada Generation NVIDIA	48 GB	Workable	BF16 everything resident	15.86 GB	4 min 56 s
GeForce RTX 4090 NVIDIA	24 GB	Slow	BF16 everything resident	15.86 GB	5 min 28 s
GeForce RTX 5090 Laptop NVIDIA	24 GB	Slow	BF16 everything resident	15.86 GB	6 min 56 s
NVIDIA DGX Spark 128GB NVIDIA	128 GB	Slow	BF16 everything resident	15.86 GB	7 min 12 s
GeForce RTX 5080 NVIDIA	16 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	8 min 0 s
GeForce RTX 4080 SUPER NVIDIA	16 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	8 min 39 s
GeForce RTX 4090 Laptop NVIDIA	16 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	9 min 0 s
GeForce RTX 4080 NVIDIA	16 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	9 min 14 s
GeForce RTX 5080 Laptop NVIDIA	16 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	9 min 29 s
GeForce RTX 5070 Ti NVIDIA	16 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	10 min 10 s
GeForce RTX 4070 Ti SUPER NVIDIA	16 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	10 min 14 s
GeForce RTX 4070 Ti NVIDIA	12 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	11 min 15 s
GeForce RTX 3090 Ti NVIDIA	24 GB	Slow	BF16 everything resident	15.86 GB	11 min 15 s
RTX A6000 NVIDIA	48 GB	Slow	BF16 everything resident	15.86 GB	11 min 37 s
GeForce RTX 4080 Laptop NVIDIA	12 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	12 min 9 s
GeForce RTX 3090 NVIDIA	24 GB	Slow	BF16 everything resident	15.86 GB	12 min 40 s
GeForce RTX 4070 SUPER NVIDIA	12 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	12 min 46 s
GeForce RTX 3080 Ti NVIDIA	12 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	13 min 14 s
GeForce RTX 5070 NVIDIA	12 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	14 min 37 s
GeForce RTX 3080 12GB NVIDIA	12 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	15 min 7 s
GeForce RTX 3080 10GB NVIDIA	10 GB	Slow	GGUF Q5_K_M everything resident	7.34 GB	15 min 7 s
GeForce RTX 4070 NVIDIA	12 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	15 min 30 s
RTX A5000 NVIDIA	24 GB	Slow	BF16 everything resident	15.86 GB	16 min 12 s
GeForce RTX 2080 Ti NVIDIA	11 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	16 min 39 s
GeForce RTX 5060 Ti 16GB NVIDIA	16 GB	Slow	GGUF Q8_0 everything resident	9.66 GB	18 min 44 s
GeForce RTX 5060 Ti 8GB NVIDIA	8 GB	Slow	GGUF Q4_K_M everything resident	6.63 GB	18 min 44 s
GeForce RTX 4070 Laptop NVIDIA	8 GB	Slow	GGUF Q4_K_M everything resident	6.63 GB	19 min 20 s
GeForce RTX 4060 Ti 16GB NVIDIA	16 GB	Barely usable	GGUF Q8_0 everything resident	9.66 GB	20 min 26 s
GeForce RTX 4060 Ti 8GB NVIDIA	8 GB	Barely usable	GGUF Q4_K_M everything resident	6.63 GB	20 min 26 s
GeForce RTX 3070 Ti NVIDIA	8 GB	Barely usable	GGUF Q4_K_M everything resident	6.63 GB	20 min 40 s
Jetson AGX Orin 32GB NVIDIA	32 GB	Barely usable	BF16 everything resident	15.86 GB	21 min 9 s
Jetson AGX Orin 64GB NVIDIA	64 GB	Barely usable	BF16 everything resident	15.86 GB	21 min 9 s
Radeon RX 7900 XTX AMD	24 GB	Barely usable	BF16 everything resident	15.86 GB	21 min 19 s
Radeon PRO W7900 AMD	48 GB	Barely usable	BF16 everything resident	15.86 GB	21 min 29 s
GeForce RTX 3070 NVIDIA	8 GB	Barely usable	GGUF Q4_K_M everything resident	6.63 GB	22 min 11 s
GeForce RTX 5060 NVIDIA	8 GB	Barely usable	GGUF Q4_K_M everything resident	6.63 GB	23 min 20 s
RTX A4000 NVIDIA	16 GB	Barely usable	GGUF Q8_0 everything resident	9.66 GB	23 min 20 s
Radeon RX 7900 XT AMD	20 GB	Barely usable	BF16 everything resident	15.86 GB	25 min 27 s
Radeon RX 9070 XT AMD	16 GB	Barely usable	GGUF Q8_0 everything resident	9.66 GB	27 min 1 s
GeForce RTX 4060 Laptop NVIDIA	8 GB	Barely usable	GGUF Q4_K_M everything resident	6.63 GB	27 min 14 s
GeForce RTX 3060 Ti NVIDIA	8 GB	Barely usable	GGUF Q4_K_M everything resident	6.63 GB	27 min 39 s
Radeon RX 7900 GRE AMD	16 GB	Barely usable	GGUF Q8_0 everything resident	9.66 GB	28 min 29 s
GeForce RTX 4060 NVIDIA	8 GB	Barely usable	GGUF Q4_K_M everything resident	6.63 GB	29 min 27 s

Source: Wan-AI/Wan2.1-T2V-1.3B-Diffusers . Downloaded 0.2 million times in the last month. See how we calculate, or browse every video model.