by Alibaba/Qwen· 8 months ago
The successor to Wan 2.1, this open-source model introduces a Mixture-of-Experts flow-matching architecture with approximately 27 billion total parameters and 14 billion active during inference. Wan 2.2 delivers significantly improved motion quality, fine-grained detail, and extended generation lengths while maintaining the accessibility of fully open weights.
Input Price
Free/M tokens
Output Price
Free/M tokens
Performance Profile
Solid video generation quality at a practical price point
Generate clips directly from text descriptions — no video editing skills required
Open source — customize for specific visual styles with LoRA fine-tuning
~27B (14B active) parameter architecture for high-fidelity generation
vs similar-tier models
| Model | Input | Output | Context | Avg Score |
|---|---|---|---|---|
Wan 2.2Current Alibaba/Qwen | Free | Free | N/A | 0.0 |
o3-mini OpenAI | $1.10 | $4.40 | 200K | 86.3 |
DeepSeek-R1 DeepSeek | $0.55 | $2.19 | 128K | 87.0 |
Generate a 5-second clip
<$0.001Short animated clip from text prompt
200 in · 5,000 out
10-second social video
<$0.001Instagram Reel or TikTok-style content
400 in · 10,000 out
Batch of 10 short clips
<$0.001Multiple variations for A/B testing
2,000 in · 50,000 out
50 ad clips per campaign
<$0.001Full video ad campaign production
10,000 in · 250,000 out
Short clips
$0.00/mo
$0.00/day
Social videos
$0.00/mo
$0.00/day
Ad production
$0.00/mo
$0.00/day
No ratings yet. Be the first to rate this model!
Sign in to rate this model and share your experience.
Sign in to leave a comment and join the discussion.
Alibaba/Qwen
Specialized code model trained on 7.5T tokens (70% code). Supports 100+ programming languages and agentic workflows.
Input
$0.30/M
Output
$0.60/M
Context
262K
Alibaba/Qwen
Most capable open VLM rivaling GPT-5 across multimodal benchmarks. Strong reasoning and agentic capabilities.
Input
$0.30/M
Output
$0.60/M
Context
128K
Alibaba/Qwen
Alibaba's open-weight hybrid MoE model with 512 experts and 17B active parameters. Natively multimodal with 201 language support. Top scores on GPQA and SWE-bench.
Input
$0.15/M
Output
$1.00/M
Context
256K
OpenAI
OpenAI's efficient reasoning model, optimized for speed while maintaining strong analytical capabilities.
Input
$1.10/M
Output
$4.40/M
Context
200K
DeepSeek
DeepSeek's reasoning model with transparent chain-of-thought. Open-source and highly competitive.
Input
$0.55/M
Output
$2.19/M
Context
128K
Anthropic
Anthropic's best balance of intelligence and speed. Excellent for production workloads.
Input
$3.00/M
Output
$15.00/M
Context
200K