Wan

What is Wan

Wan is an open-source AI video generation model series developed by Alibaba Group's Tongyi Lab. The Wan family represents Alibaba's flagship effort in multimodal AI, designed to transform text prompts, images, and reference videos into high-quality video content with realistic motion and visual consistency.

Wan 2.7, released in April 2026, expands significantly from a generation-focused model toward a broader video production toolkit — adding frame-level control, instruction-based video editing, and a built-in reasoning layer. Whether you are a marketer prototyping ad concepts or a filmmaker pre-visualizing scenes, this version introduces practical tools that address real production bottlenecks. You can select from current and legacy Wan versions in the left panel on Somake AI.

Current Version: Wan 2.7 (April 2026)

Quick Overview

Wan 2.7 — Latest Updates

Wan 2.7 expands the Wan series from a standard generation model into a more comprehensive video production suite. This release introduces frame-level precision, the ability to edit existing footage with text instructions, and an internal reasoning engine for more accurate compositions.

Key improvements in Wan 2.7:

• First-and-Last-Frame Generation: Specify both a starting frame and an ending frame. The model calculates and generates the smooth motion and transition required to connect the two logically.

• Video-to-Video Editing: Upload an existing video clip and use natural text instructions to modify style, environment, or lighting while preserving the original motion structure.

• 9-Grid Image-to-Video: Transform a 3×3 grid of storyboard shots into a single, continuous, and coherent video narrative.

• Chain-of-Thought Reasoning (Thinking Mode): A built-in reasoning layer ensures more accurate prompt interpretation, stronger compositional structure, and legible text rendering across 12 languages.

• Improved Subject Referencing: Better identity locking for consistent character and object preservation across frames. (Note: Some advanced multi-reference features may not yet be available on all third-party platforms.)

Previous version highlights (Wan 2.6):

The following features were introduced in Wan 2.6 (December 2025) and remain part of the model's capabilities:

• Multi-character dialogue with clean speech separation

• Character reference system (up to 3 references) — not yet available on Somake

• Personal avatar creation from face and voice samples — not yet available on Somake

• Environment and wardrobe control through prompts

• Native audio-visual synchronization

Objective Pros & Cons

✅ Strengths (as of v2.7):

• Thinking Mode produces noticeably more accurate compositions and readable text

• 9-grid storyboard input bridges planning and production workflows

• Remains open-source, giving developers flexibility for custom deployment

• Native audio-visual sync carried over from 2.5/2.6

⚠️ Limitations (as of v2.7):

• Voice realism and emotive audio still trail behind premium closed-source competitors like Veo 3 and Sora 2

• Complex action sequences with multiple characters may still produce visual artifacts

• Anime-style video generation remains weaker compared to realistic styles — this has not been meaningfully addressed since Wan 2.6

• Some language mismatches in output can still occur, though less frequently than in prior versions

Best Use Cases

How It Compares: Wan 2.7 vs Competitors

Who wins where:

• Wan 2.7 is the strongest choice for users who need open-source flexibility, frame-level control, and video editing capabilities at a lower cost.

• Veo 3.1 leads in audio realism and resolution — the better option if voice quality is critical and budget is not a constraint.

• Sora 2 offers longer clips and natural audio, suitable for extended narrative sequences.

You can compare Wan 2.7 against these and other models like Kling and Seedance side-by-side on Somake AI without switching platforms.

Advanced Prompting for Wan 2.6

Multi-Shot Storytelling Prompt Template

A cinematic [genre] scene.
Shot 1: [Wide/Medium/Close-up] shot, [describe scene, character, and action].
Shot 2: [Camera angle], [describe transition and new focus].
Shot 3: [Camera angle], [describe resolution or final moment].
Style: [realistic/cinematic/stylized]. Lighting: [natural/dramatic/soft].

First/Last Frame Prompt Template

A cinematic [genre] scene.
Start Frame: [Describe the starting composition, character, and lighting].
Action/Transition: [Describe the motion, camera movement, and interaction].
End Frame: [Describe the final destination].
Style: [realistic/cinematic/stylized]. Text rendering: "[Include any specific text to appear on screen in quotes]".

Prompt Expansion

Enable prompt expansion when your input is simple or you want richer visual detail. The system adds descriptive elements to improve composition, style consistency, and visual coherence in the output.

Troubleshooting Common Issues

Problem: Voice sounds robotic or unnatural
→ Solution: This remains a limitation as of Wan 2.7. For projects requiring highly realistic or emotive voices, consider using the video output with separately generated or recorded audio.

Problem: Unexpected characters or surreal elements appear
→ Solution: While Wan 2.7's Thinking Mode reduces these artifacts, they can still occur with overloaded prompts. Simplify your prompt and regenerate.

Problem: Action scenes have visual distortions
→ Solution: Complex action sequences with multiple characters are a known weakness. Break dynamic scenes into simpler shots and use the first-and-last-frame feature to explicitly control the start and end points of motion.

Problem: Text on signs or clothing is garbled
→ Solution: Wan 2.7 has improved text rendering across 12 languages via Thinking Mode. If text is still garbled, explicitly wrap the desired text in quotes in your prompt and avoid blending contradictory reference images.