FrameCost
Beta. Estimates only. Timings come from the model makers' own published benchmarks, and prices from provider pricing pages and APIs. Your settings, drivers and cloud host will change real costs. We have run no benchmarks ourselves.
Home › Models › Wan2.2 T2V-A14B › VRAM

Wan2.2 T2V-A14B VRAM requirements

In the official benchmark, Wan2.2 T2V-A14B peaked at 26.2 GB per GPU (T2V 480P on 8× H20), the lowest of 18 measured settings. README: README note on the single-GPU commands: "This command can run on a GPU with at least 80GB VRAM." (The benchmark single-GPU runs use model offloading.).

Last checked 2026-10-17 on the official pages.

Peak GPU memory per GPU from the model maker's own benchmark table and the README. We have not run these ourselves.

SettingGPUsPeak memory / GPUSource
T2V 480P8× H2026.2 GBbenchmark · 2026-10-17
T2V 480P8× A100/A80026.3 GBbenchmark · 2026-10-17
T2V 480P8× H100/H80026.3 GBbenchmark · 2026-10-17
T2V 720P8× H2037.0 GBbenchmark · 2026-10-17
T2V 720P8× A100/A80037.0 GBbenchmark · 2026-10-17
T2V 720P8× H100/H80037.1 GBbenchmark · 2026-10-17
T2V 480P4× H2040.9 GBbenchmark · 2026-10-17
T2V 480P4× A100/A80040.9 GBbenchmark · 2026-10-17
T2V 480P4× H100/H80040.9 GBbenchmark · 2026-10-17
T2V 480P1× H2041.3 GBbenchmark · 2026-10-17
T2V 480P1× A100/A80041.3 GBbenchmark · 2026-10-17
T2V 480P1× H100/H80041.3 GBbenchmark · 2026-10-17
T2V 720P4× H2051.7 GBbenchmark · 2026-10-17
T2V 720P4× A100/A80051.7 GBbenchmark · 2026-10-17
T2V 720P4× H100/H80051.7 GBbenchmark · 2026-10-17
T2V 720P1× H2059.8 GBbenchmark · 2026-10-17
T2V 720P1× A100/A80059.8 GBbenchmark · 2026-10-17
T2V 720P1× H100/H80059.8 GBbenchmark · 2026-10-17

Stated in the README

Tracked cloud GPUs with at least 26.2 GB

OfferVRAMPrice
Vast.ai · A100 PCIE80 GB$1.00/hr
Runpod · A100 PCIe80 GB$1.19/hr
Vast.ai · A100 SXM480 GB$1.20/hr
Runpod · A100 SXM80 GB$1.39/hr
Runpod · A100 SXM80 GB$1.79/hr
Runpod · A100 PCIe80 GB$1.79/hr
Runpod · H100 PCIe80 GB$1.99/hr
Vast.ai · H100 PCIE80 GB$2.16/hr
Runpod · H100 SXM80 GB$2.69/hr
Runpod · H100 PCIe80 GB$2.89/hr
Vast.ai · H100 SXM80 GB$3.26/hr
Runpod · H100 SXM80 GB$3.99/hr

Rent on Runpod ↗ · Rent on Vast.ai ↗

Disclosure: the Runpod and Vast.ai sign-up links are referral links, so we may earn credit or a commission if you sign up through them. Costs and the cheapest-first ordering are calculated identically for every provider and never depend on referral status.

Wan2.2 T2V-A14B cost per video · generation time

Frequently asked questions

How much VRAM does Wan2.2 T2V-A14B need?

In the official benchmark, Wan2.2 T2V-A14B peaked at 26.2 GB per GPU (T2V 480P on 8× H20), the lowest of 18 measured settings. README: README note on the single-GPU commands: "This command can run on a GPU with at least 80GB VRAM." (The benchmark single-GPU runs use model offloading.).

What is the cheapest cloud GPU with enough VRAM for Wan2.2 T2V-A14B?

Vast.ai A100 PCIE (80 GB) at $1.00/GPU-hr meets the lowest measured peak of 26.2 GB. Peak memory can differ with your settings.