OpenArt promotes Alibaba's Qwen-Image-3.0 while its API remains in limited preview
Coco Mao and John Qiao's creator platform is promoting Alibaba's image model, which targets dense layouts and multilingual text while its cloud API remains in limited preview.
By RuntimeWire Staff · Published
Primary source: Qwen / Alibaba
Why it matters
Qwen-Image-3.0 targets a persistent weakness in image generation: producing dense layouts with small, multilingual text. OpenArt gives the model exposure to a large creator audience, while developers seeking direct cloud access must apply for Alibaba's limited preview.
OpenArt, the San Francisco AI creative platform founded by Coco Mao and John Qiao, is promoting Alibaba's Qwen-Image-3.0 following the model's July 21 launch. The release is aimed at a stubborn problem for image generators: producing information-dense layouts with small, multilingual text.
Alibaba's launch materials say Qwen-Image-3.0 accepts inputs of up to 4,500 tokens, renders text in 12 languages and can generate formats including newspapers, storyboards, exam papers and interface mockups. Its launch demonstrations claim legible text at roughly 10 pixels. Those performance claims come from Alibaba's launch materials and have not been independently tested in the reviewed sources.
OpenArt has a promotional page for Qwen Image 3.0. A post from Qwen's official X account also directs users to try the model on OpenArt and links to an OpenArt post.
The reviewed materials do not specify the date or commercial terms of the OpenArt arrangement. They also do not establish an investment, acquisition, exclusive agreement or broader partnership between Alibaba and OpenArt.
Long prompts for information-dense images
Qwen-Image-3.0 is designed to handle prompts and compositions that strain conventional image generators. Alibaba's launch materials show a nine-panel educational graphic described with a 3,700-token prompt, along with examples involving mathematical notation, nested software interfaces, newspapers and academic papers.
Alibaba attributes that output to a 4,500-token input limit and improved spatial control. Alibaba also claims the model can reproduce fine details such as hair, skin texture and annotations while rendering text as small as 10 pixels. The examples demonstrate Alibaba's intended use cases, although Alibaba has not supplied independent benchmark results in the reviewed materials.
Alibaba Cloud identifies the hosted model as qwen-image-3.0-pro. The API supports text-to-image generation and editing with one to three reference images, with output resolutions up to 2048 x 2048 pixels. The documentation was updated July 20 and says prospective users must apply for limited-preview access through the Model Gallery.
That access requirement limits immediate API adoption by developers. It also leaves pricing, capacity and general availability unresolved in the reviewed sources. Alibaba's launch post directs users to Qwen Chat, while its cloud documentation provides the separate application-only API route.
OpenArt brings Qwen to a large creator audience
OpenArt aggregates third-party image and video models alongside editing, character-consistency and storytelling tools. Mao and Qiao started the company in 2022 after leaving Google, where they had built a short-form video product inside Area 120. According to OpenArt's account of its history, the five-person project reached millions of monthly active users before Google Search acquired it.
The founders initially paired generated images with the prompts used to create them, an idea they described as "Pinterest for AI images." Requests for consistent characters later pushed the product toward editing, storytelling and persistent creative worlds.
OpenArt has enough scale to attract model providers. On January 30, Canaan said OpenArt had reached 8 million monthly active users and more than $70 million in annual recurring revenue after growing revenue sevenfold in 2025. Canaan said OpenArt had roughly 20 employees at the time.
Canaan announced that it led a $30 million Series A with Basis Set Ventures and DCM participating. OpenArt's company history describes the financing as $35 million, leaving a $5 million discrepancy between the founders' account and the investors' announcements.
Qwen Chat is the public route identified in Alibaba's July 21 launch announcement. Alibaba Cloud offers a separate API route under limited preview and requires an application through the Model Gallery. OpenArt's promotion gives Qwen-Image-3.0 exposure to its creator audience, while the model's appeal to engineers and creative-tool companies will depend on whether its claimed gains in typography, layouts and reference-image editing survive outside Alibaba's demonstrations, and when Alibaba opens broader API access.