Wan 2.6 Flash

• 🎬 Alibaba's fast image-to-video model, converting a still into motion
• 📏 Outputs 720p or 1080p video from a first-frame image
• 💬 Optional native audio with lip-sync and ambient sound
• 🎯 Aims for strong subject coherence across generated frames
• 🎥 Supports single continuous shots or automatic multi-shot transitions
• 🔧 Accessed via DashScope/Model Studio as wan2.6-i2v-flash

Alibaba

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,…

Wan 2.6 Flash is the speed-optimized image-to-video member of Alibaba's Wan 2.6 family, taking a single still image plus a text prompt and animating it into a smooth, photorealistic clip. It is offered through Alibaba's Model Studio as the endpoint wan2.6-i2v-flash, producing 720p or 1080p output and supporting optional native audio, which can be disabled for silent clips. A distinguishing capability of the 2.6 generation is automatic multi-shot narrative, keeping the subject consistent across shot transitions.

The Flash designation marks the latency-focused tier of the line, intended for rapid prototyping and previews while aiming to retain the motion coherence and subject consistency that characterize the series.

Within the family, it sits alongside the full Wan 2.6 image-to-video model released the prior month, and follows the earlier Wan 2.5 Preview. Alibaba's later Wan 2.7 documentation notes that 2.6 and earlier support only first-frame-to-video, whereas 2.7 adds first-and-last-frame and video continuation. Flash trades some of the full model's headroom for the lower latency suited to quick iteration.