Wan 2.6
The most open AI video model — 14 billion parameters, Apache 2.0 license, 92% character consistency, and available on Flashloop's free tier.
Max Duration
15 seconds
Resolution
720p
Aspect Ratios
16:9, 9:16, 1:1
Audio
Native audio generation
What makes Wan 2.6 unique
Open source under Apache 2.0 — weights, training code, and inference scripts all available publicly
14 billion parameter Diffusion Transformer with Mixture-of-Experts design
92% character consistency across 8+ shots — the highest reported figure of any model
Reference-to-video: upload a character with appearance and voice to generate new scenes starring them
Available on Flashloop's free tier — no subscription required
Can run locally on consumer GPUs (720p on 24GB VRAM) thanks to 64x compression VAE
Best for
How Wan 2.6 works
Wan 2.6 is a 14 billion parameter Diffusion Transformer with a Mixture-of-Experts (MoE) design. Dual expert networks handle different stages: a high-noise expert for layout and structure, and a low-noise expert for detail refinement. A high-compression VAE with 64x compression ratio enables it to run on consumer GPUs. Trained on 1.5 billion videos and 10 billion images.
How to use Wan 2.6 on Flashloop
Select Wan 2.6
Open Flashloop and choose Wan 2.6 from the model selector in the video creator.
Enter your prompt
Describe what you want to see in your video — be as detailed as you like.
Generate & download
Hit generate and your video will be ready in seconds. Download or share directly.
Ready to create with Wan 2.6?
Start generating videos with Wan 2.6 on Flashloop — available on iOS, Android, and web.
Start Creating with Wan 2.6