Wan 2.7, released April 3 2026 by Alibaba’s Tongyi Lab, is the current flagship of the Wan video series. It generates 1080p clips with native audio sync (dialogue, music and sound effects), and adds professional controls — first/last-frame anchoring, 9-grid multi-scene input and subject + voice cloning — at roughly $6 per minute, about a third cheaper than Wan 2.5/2.6.
Wan matters as much for its lineage as its features: Wan 2.1 and 2.2 were open-sourced under Apache-2.0 and became a default base for the self-hosted ComfyUI community. That history is the main reason the series is the go-to Chinese open-video ecosystem.
The honest caveat: the open era paused with 2.2. Versions 2.5 onward launched as commercial cloud APIs, and while the community expects open weights for 2.7 to follow Alibaba’s usual cloud-first pattern, that hasn’t been confirmed. It also rewards structured prompting more than plug-and-play tools do.
Best for
- 1080p clips with synced audio and multiple shots
- Reference-driven, character-consistent storytelling
- Self-hosting via the open Wan 2.1/2.2 base
Pros & cons
Strengths
- Native audio-visual sync (dialogue, music, SFX)
- First/last-frame anchoring and 9-grid multi-scene input
- Subject + voice cloning; strong open-source lineage
Limitations
- Wan 2.5+ are cloud-only; 2.7 open weights not yet confirmed
- Steeper, structured-prompting learning curve