How much VRAM does Wan 2.2 TI2V-5B need?
Wan 2.2 TI2V-5B needs about 6.9 GB at 8-bit, or 4.8 GB as a 4-bit GGUF, at its default output size.
- Type
- Video model
- Parameters
- 5B
- Text encoder
- UMT5-XXL
- Released
- Jul 2025
- License
- Apache 2.0
Memory at each precision
| Precision | Weights | Peak at 704p, 5 s |
|---|---|---|
| 16-bit BF16 | 9.3 GB | 11.4 GB |
| 8-bit FP8 · Q8_0 | 4.8 GB | 6.9 GB |
| 6-bit GGUF Q6_K | 3.8 GB | 5.9 GB |
| 5-bit GGUF Q5_K_M | 3.3 GB | 5.4 GB |
| 4-bit GGUF Q4_K_M · NVFP4 | 2.8 GB | 4.8 GB |
The UMT5-XXL text encoder (5.3 GB at FP8) runs first and then waits in system RAM.
What runs it
| Machine | Best fit | Per clip |
|---|---|---|
| RTX 3060 12 GB | 16-bit | 54 min |
| RTX 4060 Ti 16 GB | 16-bit | 31 min |
| RTX 3090 | 16-bit | 20 min |
| RTX 4090 | 16-bit | 8.6 min |
| RTX 5090 | 16-bit | 6.9 min |
| RTX PRO 6000 Blackwell | 16-bit | 3.1 min |
| M4 Max 128 GB | 16-bit | 90 min |
| M5 Max 128 GB | 16-bit | 26 min |
| Ryzen AI Max+ 395 128 GB | 16-bit | 40 min |
| DGX Spark 128 GB | 16-bit | 22 min |
| M3 Ultra 256 GB | 16-bit | 50 min |
| M5 Ultra 512 GB | 16-bit | 16 min |
| H100 SXM | 16-bit | 1.7 min |
| DGX Station GB300 | 16-bit | 53 s |
704p, 5 s, 50 steps. Times are rough, plus or minus 30%.
Where to get it
Questions
How much VRAM does Wan 2.2 TI2V-5B need?
Wan 2.2 TI2V-5B needs about 6.9 GB at 8-bit, or 4.8 GB as a 4-bit GGUF, at its default output size.
Can Wan 2.2 TI2V-5B run on an RTX 4090?
Yes. At 16-bit it takes about 8.6 min per clip.
More models
Data checked Oct 2026.