Alibaba's Wan video generation model may be getting its next update sooner than expected. A post on the Russian-language Telegram channel @cgevent claims Wan 2.7 will arrive this month with a batch of new capabilities, though Alibaba has made no official announcement.
The rumored feature list includes first-frame and last-frame video generation (a capability already present in Wan 2.1 and 2.2), 9-grid image-to-video, subject and voice reference, instruction-based video editing, and video recreation. Most of these build on existing Wan pipelines. The official site still lists Wan 2.6 as the current release, which introduced multi-shot storytelling, reference-driven generation, and native audio sync up to 15 seconds at 1080p.
The biggest claim: no code. If accurate, Wan 2.7 would continue the shift that started with Wan 2.5, which broke from Alibaba's open-source tradition by offering API-only access. Earlier versions like Wan 2.2 shipped under Apache 2.0 on GitHub and became a favorite among local-deployment enthusiasts running consumer GPUs.
None of these details have been confirmed by Alibaba or the Wan AI team. The @cgevent channel covers AI and metaverse news but isn't an official source. March is ticking, so if Wan 2.7 is real, some kind of announcement should surface soon.
Bottom Line
An unverified Telegram post claims Wan 2.7 launches this month with video editing features but no open-source release, continuing Alibaba's shift toward closed APIs.
Quick Facts
- Current version: Wan 2.6 (per wan.video)
- Rumored features: first/last frame generation, 9-grid I2V, voice reference, video editing, video recreation
- No code release expected (unverified claim)
- Source: @cgevent Telegram channel (not official)
- Wan 2.5 was Alibaba's first closed-source Wan release
