Alibaba Wan3.0: A 30-Second AI Video Tool Targets Marketing and Corporate Comms with Closed Weights and Staged Access

Alibaba has launched Wan3.0, its latest video generation model that can turn PDFs, slide decks, and web pages into 30-second clips at up to 1080p, priced well below Google Veo 3.1. For AI builders evaluating video generation tooling, the closed weights and staged access mark a clear shift from Alibaba's historically open Wan series toward a more commercial, controlled deployment model.

Document-to-video for marketing and corporate teams

The most distinctive capability of Wan3.0 is what it accepts as input. Alongside text, images, audio, and video, the model parses PDFs and PowerPoint files, converting structured documents directly into video content that mirrors the source material's layout. Alibaba targets marketing departments, corporate communications, short drama production, and even simulation training for autonomous vehicles and robotics. The model maintains character consistency, props, spatial layout, and motion graphics across the full 30-second duration, with synchronized facial expressions and multilingual voice output. This positions Wan3.0 as a business workflow tool rather than a creative filmmaking tool, though Alibaba also lists social content and music video creation as use cases.

Pricing that undercuts Veo but faces its own pressure

Wan3.0 runs at $0.05 per second at 480p, $0.10 at 720p, and $0.20 at 1080p, which works out to about $12 per minute at the highest tier. That is significantly cheaper than Google Veo 3.1's standard tier at $0.40 per second. However, the price comparison is not one-sided: several rivals that outrank Alibaba's previous Wan2.2 on independent leaderboards cost less per minute than Wan3.0 at 1080p, and Wan3.0 itself has not yet appeared on any independent benchmark. An Indian startup recently shipped a video model at half a cent per second, betting that most commercial video does not need frontier quality. The pricing landscape is shifting fast, and Wan3.0's actual quality relative to cost remains a vendor claim for now.

Closed weights and staged access change the developer calculus

Unlike Alibaba's previous Wan series models, which were open-sourced (Wan2.2 was the last open release in July 2025), Wan3.0 ships with closed weights. Access is staged: public beta runs through Alibaba Cloud Model Studio and Qwen Cloud on an application basis, full API availability is described as rolling out, and a consumer-facing site at wan.video is planned as a members-only platform. This suggests Alibaba is metering demand while it calibrates serving costs, and it means developers cannot self-host or fine-tune the model. For builders integrating video generation into products, this limits flexibility compared to open alternatives and introduces dependency on Alibaba's API pricing and availability.

What remains uncertain

Without independent benchmarks, the competitive position is unclear. Alibaba's quality claims cannot be verified externally, and several lower-cost competitors already exist. The closed weights also rule out community scrutiny and customization. On the business side, Alibaba's Cloud and AI revenue grew 45% year-over-year to 48.44 billion yuan in the June quarter, but quarterly net profit fell 75% as capital expenditure rose 75% to 67.68 billion yuan. The company committed 380 billion yuan to AI infrastructure over three years. Whether Wan3.0 justifies that spend is an open question, but for builders, the near-term decision hinges on whether the document-to-video workflow and pricing beat the alternatives in their specific use case.