Wan 2.2 TI2V 5B.
Wan 2.2 TI2V 5B is an open-source video model from Alibaba (Wan-AI). Windpaint doesn't serve it yet, so there's no model id or price. Request early access to start with the live models.
- Status
- Coming soon
- Task
- Text to video, Image to video
- Input
- Prompt, optional start frame
- Output
- Clip at 720p, 24 fps
- Developer
- Alibaba (Wan-AI)
- Parameters
- 5B
- Licence
- Apache-2.0
- Released
- July 2025
What the model does.
Paraphrased from the official model card.
Wan2.2-TI2V-5B is the compact model in the Wan 2.2 release. One 5B dense model does both text-to-video and image-to-video at 720p and 24 frames per second.
It uses the new Wan 2.2 VAE, which compresses video more heavily than the larger models, and its README says it runs on a single consumer GPU such as an RTX 4090. For this model, 720p means 1280 × 704 or 704 × 1280.
Coming soon to Windpaint.
Wan 2.2 TI2V 5B isn't served, so it has no model id, no price and no request example here. When it goes live, this page gets its tiers, prices and request fields like the live models have.
Want this model? Request early access. We're adding models as we add capacity.
Is Wan 2.2 TI2V 5B licensed for commercial use?
Licence: Apache-2.0.
Yes. Apache-2.0 allows commercial use, including running the model behind a paid API, with no revenue or user caps.
- Apache 2.0. The Wan team states it claims no rights over generated content. Wan2.2 repo
This is our reading of the licence, not legal advice. Read the licence text before you rely on it.
Wan 2.2 TI2V 5B VRAM requirements.
Published by Alibaba (Wan-AI) for running the weights yourself. These are the developer's figures, not Windpaint's.
- Published VRAM
- At least 24 GB
- GPUs named
- RTX 4090
- Applies to
- Single-GPU 720p command with model offload, dtype conversion and T5 on CPU. Drop those options at 80 GB or more.
- The README's single-GPU TI2V-5B command (
--offload_model True --convert_model_dtype --t5_cpu) can run on a GPU with at least 24GB VRAM (e.g. RTX 4090 GPU). Wan2.2 repo - On a GPU with at least 80GB VRAM, the README says you can remove
--offload_model True,--convert_model_dtypeand--t5_cpu. Wan2.2 repo - Multi-GPU inference uses FSDP plus DeepSpeed Ulysses; the official example uses 8 GPUs. Wan2.2 repo
- Diffusers-format weights are published as
Wan-AI/. The README also lists ComfyUI integration and a Hugging Face Space running TI2V-5B. Wan2.2 repoWan2.2-TI2V-5B-Diffusers
Wan 2.2 TI2V 5B native resolution.
What the developer's own code and examples use.
- 720P only, and for this model 720P means 1280 × 704 or 704 × 1280, not 1280 × 720. Wan2.2 repo
- Config defaults:
frame_num=121,sample_fps=24. Wan2.2 repo, wan_ti2v_5B.py - With an image,
--sizesets the output area and the aspect ratio follows the input image. Wan2.2 repo - The Wan2.2-VAE compresses 4×16×16 (T×H×W); with an additional patchification layer, TI2V-5B's total compression reaches 4×32×32. Wan2.2 repo
- The A14B configs use the Wan2.1 VAE with stride (4, 8, 8); TI2V-5B uses the Wan2.2 VAE with stride (4, 16, 16). Wan2.2 repo, wan_t2v_A14B.py
Recommended settings for Wan 2.2 TI2V 5B.
Steps, guidance and defaults from the developer's card and code, for running the weights yourself.
- Official config: 50 sampling steps, guidance scale 5.0, sample_shift 5.0, 121 frames, 24 fps. Wan2.2 repo, wan_ti2v_5B.py
- Diffusers example: 1280 × 704,
num_frames=121,num_inference_steps=50,guidance_scale=5.0, exported atfps=24. Wan2.2-TI2V-5B-Diffusers model card
Wan 2.2 TI2V 5B prompt guide.
From Alibaba (Wan-AI)'s own prompt guide, prompt tooling and examples.
- The README's TI2V section points to the same prompt-extension process used for the A14B models. Wan2.2 repo
- Passing
--imagemakes it image-to-video; without an image it defaults to text-to-video. Wan2.2 repo
Wan 2.2 TI2V 5B limitations.
As the developer states them.
- Only two sizes are accepted for ti2v-5B:
1280*704and704*1280. There is no 480P option. Wan2.2 repo, __init__.py generate.pysays the frame count should be 4n+1. Wan2.2 repo, generate.py
Questions.
Can I call Wan 2.2 TI2V 5B on Windpaint today?
No. Wan 2.2 TI2V 5B isn't served yet, so it has no model id, no price and no entry in the catalog endpoint. There's no date for Wan 2.2 TI2V 5B.
How much VRAM does Wan 2.2 5B need?
The README says its single-GPU command, with model offloading, dtype conversion and the T5 encoder on CPU, can run on a GPU with at least 24GB of VRAM, such as an RTX 4090. On a GPU with 80GB or more you can remove those options.
What is the native resolution of Wan 2.2 5B?
720p, which for this model means 1280x704 or 704x1280. It generates at 24 fps.
Can Wan 2.2 5B do image to video?
Yes. One model does both text-to-video and image-to-video. Pass an image and it runs image-to-video; leave it out and it runs text-to-video.
Wan 2.2 5B vs 14B: what's the difference?
TI2V-5B is a 5B dense model that does text-to-video and image-to-video in one, at 720p and 24 fps, with the high-compression Wan2.2-VAE. The A14B models are 27B mixture-of-experts models with 14B active per step, released as separate T2V and I2V checkpoints that support 480p and 720p.
Is Wan 2.2 TI2V 5B open source?
Yes. Its weights are licensed Apache-2.0, an OSI-approved open-source licence.
Can Wan 2.2 TI2V 5B be used commercially?
Yes. Apache-2.0 allows commercial use, including running the model behind a paid API, with no revenue or user caps. Its licence is Apache-2.0. That's our reading of the licence, not legal advice; check the licence text before you rely on it.
How do I ask for Wan 2.2 TI2V 5B?
Request early access at app.windpaint.ai/register. We're letting people in gradually while we add models and capacity.