AlibabaAlibaba·🎬 Video Generation·↑ Newer: Wan 2.7

Wan 2.5 Preview

anonymized
Try on Venice.ai ↗
Quick reference
Wan 2.5 Preview — TLDR
  • 🆕 Alibaba's Wan 2.5 image-to-video preview, released September 2025.
  • 👁️ Animates a still first frame into motion video.
  • 🔧 Accepts a first frame plus an optional text prompt.
  • 🎯 Photorealistic generation with strong subject coherence across frames.
  • 🏢 Built by Alibaba's Wan video team.
  • 📚 Part of the broader Wan generative video family.
  • 🆕 Released as a preview ahead of later Wan versions.
💰 Pricing
$0.280 – $1.65
per generation
📅 On Venice since
Sep 25, 2025
296 days ago
Provider

Alibaba Group is a Chinese multinational technology company founded in 1999 and headquartered in Hangzhou, Zhejiang. Originally built around e-commerce and cloud computing, Alibaba has become one of the most prolific contributors to open-weight AI research,…

Read full profile →
51 models on Venice
20 video · 18 text · 5 image · 4 inpaint · 2 embedding · 2 tts
Since Jan 11, 2025

About this model

Wan 2.5 Preview is the image-to-video member of Alibaba's Wan generative video family, released in September 2025. It takes a static first frame plus an optional text prompt and animates it into a photorealistic clip, with an emphasis on subject coherence and stability across frames. Alibaba Cloud documents the image-to-video workflow around supplying a first frame from which motion is generated.

As the catalog notes, the model targets photorealistic generation with strong subject coherence across a sequence, positioning it as the preview-stage entry in the Wan image-to-video line.

The family has since advanced beyond this preview, with later versions released under the Wan banner. Wan 2.6 arrived in December 2025, and Wan 2.7 followed in April 2026, each succeeding the 2.5 generation.

As a preview release, Wan 2.5 represents an earlier iteration of Alibaba's image-to-video approach, with subsequent versions continuing the family's development.

This About section is AI-generated from public sources (Claude Opus 4.8), with no human editing. It may contain inaccuracies — verify critical details against the sources listed above.

Data sources: Venice API · HuggingFace · Wikipedia — enrichment updated 4d ago