Skip to content
#55BJul 28, 2025Evidence A

Wan2.2

Alibaba · 27B total / 14B active (MoE)

An Apache-2.0 base for video generation

Alibaba’s video-generation model open-sourced in July 2025: 27B-A14B, Apache-2.0, generation 94, video 86, ecosystem 88 and a composite of 65.1. It is one of the most common bases for community fine-tuning and ComfyUI workflows; understanding is weak (perception 72, documents 55), so it is not an analysis model.

TextImageVideoNative

Last updated Oct 21, 2026Editorial scoring

Bottom line

For a video-generation base with clean licensing, mature community workflows and fine-tunability, Wan2.2 remains one of the safest open options in 2026: Apache-2.0, 27B-A14B, ecosystem 88. Its low score comes from understanding dimensions, not generation quality; do not reject it on the composite alone.

Eight-dimension breakdown

editorial mapping · 0–100
Perception72Document55Chart QA40Video86Agent40Generation94Chinese82Ecosystem888D
  • Wan2.265.1
Visual perception & grounding20%
72
Document parsing & Chinese OCR18%
55
Chart & document reasoning14%
40
Video & long-horizon reasoning14%
86
GUI agents & tool use14%
40
Generation & editing6%
94
Chinese context & culture6%
82
Ecosystem, cost & availability8%
88

Weights: perception 20 / documents 18 / charts 14 / video 14 / agents 14 / generation 6 / Chinese 6 / ecosystem 8. Definitions and sources on the methodology page.

Public benchmarks

as reported, not converted
Benchmark / metricValueNote
参数规模27B-A14BMoE
许可证Apache-2.0Commercial use
上下文16KFor generation tasks

Public scores come from different versions and prompts and are not interchangeable with our composite; they are one evidence input.

Scenario mapping

five stress tests, one scale
ScenarioScoreObservation
Short-video generation94 · Stable style and camera
Image generation90 · Good first-frame quality
30-min video QA86 · Generation only, not QA
Chinese receipt OCR55 · Not an OCR model
Chinese cultural context82 · Good response to Chinese-style prompts

Strengths and limitations

Strengths

  • Apache-2.0 plus 27B-A14B is the best balance of licence and capacity among video-generation bases.
  • Ecosystem 88: the richest ComfyUI, LoRA and community workflows.
  • Generation 94 and video 86 keep camera motion and stylisation stable.

Limitations

  • Perception 72 and documents 55: not for multimodal analysis.
  • Agents 40 and charts 40: unsuitable for business automation.
  • 16K context limits long prompts and multi-shot scripts.

Price and real cost

List price

—N/A

Open weights; self-hosted, no API fee. For 1,000 images at 1,024px, ~1,100 image tokens each and 300 output tokens, the monthly cost is about ¥0.0 (assuming input at 25% of output price).

Use your volume

About Wan2.2

Is Wan2.2 open source?

Licence: Open weights (Apache-2.0). Weights are available, but check the exact terms and filing requirements before commercial use.

How much does Wan2.2 cost?

Open weights; self-hosted, no API fee. It ranks #55 with a composite of 65.1. Real bills depend on resolution, tiling, caching and volume discounts.

What is Wan2.2 best for?

Video generation and stylisation; ComfyUI workflows; A base for video-model fine-tuning。

Why does Wan2.2 score 65.1?

The composite weights eight dimensions: perception 20%, documents 18%, charts 14%, video 14%, agents 14%, generation 6%, Chinese 6%, ecosystem 8%. Its strongest dimension is Generation & editing; its weakest is Chart & document reasoning.

Personal ranking · Not an official score

Every ranking, score and comparison on this page is an independent personal ranking, not an official score, and represents no vendor, institution or regulator. Scores are a subjective mapping of public evidence under our own weights — change the weights and the order changes. Disagree, or spotted a factual error? Contact the site owner: jizhi2024@foxmail.com — we check each report and log corrections in the changelog.