Alibaba Launches Wan2.7-Video

Alibaba has unveiled Wan2.7-Video, a next-generation video generation model designed to transform creators from executors into directors by providing full creative workflow control.

The launch follows closely on the heels of its Wan2.7-Image release, marking a rapid expansion of Alibaba’s AI multimedia capabilities.

The new system allows creators to orchestrate complex narratives and post-production processes with unprecedented precision, enabling director-grade video output from a single prompt. Users can manage script input, character actions, dialogue, visual styles, and camera movements, all while maintaining consistent lighting, voice, and visual identity across multiple scenes.

Wan2.7-Video includes four specialized models: Text-to-Video (Wan2.7-t2v), Image-to-Video (Wan2.7-i2v), Reference-to-Video (Wan2.7-r2v), and Video Editing (Wan2.7-videoedit). Together, these models unify text, image, video, and audio inputs into a single platform for generation, editing, reshaping, continuation, and referencing. The system supports video lengths from 2 to 15 seconds, outputting 720p or 1080p resolution, with enterprise APIs available for batch processing and customized workflows.

The model addresses critical production challenges such as multi-shot consistency, realistic lip-syncing, and dynamic dialogue adjustments. Multimodal inputs allow users to control story elements in detail—for example, altering weather through audio cues or adjusting camera composition and character settings via reference images. The system maintains cross-video consistency for up to five distinct characters, with thousands of style combinations and over 50 refined emotional expressions for lifelike performances.

Wan2.7-Video enables creators to modify any aspect of a video using natural language instructions. "A single prompt can yield a fully realised storyboard featuring professionally constructed, multi-storyboard paced transitions, FPV drone dives, 360-degree orbital shots, and context-aware lighting," Alibaba said. The model also includes a video continuation mechanism to specify ending frames, ensuring smooth transitions without abrupt cuts.

The release of Wan2.7-Video is complemented by Wan2.7-Image, which offers highly personalised visual generation with professional-grade precision. Wan2.7-Image supports fine-tuning of character traits, exact color palette matching, and advanced text rendering across 12 languages. It allows batch processing of up to nine reference images and produces 12 outputs per batch, streamlining storyboarding and e-commerce workflows. Alibaba also launched Wan2.7-Image-Pro, featuring sharper prompt interpretation, stable composition, and 4K output.

Together, the Wan2.7 suite aims to provide creators and businesses with the tools to produce professional-quality video and imagery more efficiently, bridging the gap between creative intent and technical execution.