Wan 2.7

• 🏢 Alibaba's Wan text-to-image model, released April 2026.
• 🎯 Catalog describes photorealistic outputs with strong detail preservation.
• 🔧 Unified text-to-image generation and editing in one model.
• 🖼️ Supports interactive image editing alongside generation.
• 📚 Served via Alibaba Cloud Model Studio, billed per image.
• 🆕 Newest static-image entry in the Wan 2.7 line.
• 🔒 Input prompts and outputs pass content moderation.

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 2.7 is the text-to-image variant of Alibaba's Wan family, served through Alibaba Cloud Model Studio and developed by the Tongyi Lab. According to Alibaba's API documentation, it is a unified model that combines text-to-image generation with interactive editing, billed per successfully generated image. The catalog describes it as producing photorealistic outputs with strong detail preservation, suited to marketing assets, product visualization, and creative work.

Within the 2.7 generation, the standard model sits below Wan 2.7 Pro, the higher-tier image variant in the same line. For instruction-guided edits, the family also includes Wan 2.7 Pro Edit, aimed at detail-preserving, multi-element modifications. Alongside it, Alibaba's separate Qwen Image 2 line offers another text-to-image option in the same provider ecosystem.

Relative to earlier Wan releases, the 2.7 generation extends Alibaba's combined generation-and-editing approach documented in the Model Studio API reference, building on the prior Wan 2.6 image pipeline. Documented capabilities center on text-to-image generation together with image editing through a single endpoint.

Note that no provider-published or independent benchmark scores were available for this model at the time of writing, so this summary stays to documented features rather than performance rankings. Video output and animation remain outside this static-image variant's scope.