Alibaba is preparing to charge some large commercial users of its next open-source AI model, according to a Reuters exclusive citing people familiar with the plan. The report suggests the company is considering a revenue-sharing arrangement rather than making the model entirely free for its biggest users.

The reported shift matters because it would test a difficult business model in AI: distributing model technology openly while asking high-volume commercial customers to share revenue or pay for usage. Alibaba has not publicly confirmed the reported terms in the source material available for this story, and details such as pricing, eligibility thresholds and the model’s release date remain unclear.

What Reuters is reporting

Reuters identified the upcoming system as the next major model in Alibaba’s Qwen family. Its report describes the model as open source and says Alibaba plans to seek payment from “big users” operating commercial applications. A separate Investing.com listing characterized the proposal as revenue sharing for commercial users, while coverage from Межа. Новини. України used similar language about large users of an open Qwen model.

The three source items do not provide the operational details needed to determine how the plan would work. They do not specify whether Alibaba would charge for model downloads, hosted inference, commercial deployment, or a percentage of customer revenue. They also do not establish whether the arrangement would apply globally or only in selected markets.

That distinction is important. An open-source AI model can be downloaded and run outside the vendor’s infrastructure, but a company can still charge for managed access, support, additional services or commercial rights. The available reporting does not say which of those approaches Alibaba intends to use.

Why the Qwen decision matters

Qwen has become an important part of Alibaba’s effort to compete in the global AI market, but the evidence supplied here does not include new benchmarks, adoption figures or technical specifications for the next release. The immediate news is commercial rather than technical: Alibaba is reportedly looking for a way to monetize companies that build substantial businesses on top of its model.

For AI developers, that raises a question about what “open” will mean in practice. Open availability can reduce barriers to experimentation and allow teams to host models in their own environments. A revenue-sharing policy could preserve that access for research and smaller projects while creating a separate obligation for high-volume commercial deployment. Whether that balance is attractive will depend on the eventual license and enforcement terms.

The proposal also places Alibaba in a broader competitive debate involving model providers, cloud platforms and independent AI developers. Companies increasingly have several ways to obtain model capability: they can call a hosted API, use a managed cloud service, or download weights and operate infrastructure themselves. A charge aimed at large users could make the open Qwen model financially resemble a commercial platform for the customers most capable of switching between those options.

Evidence and unresolved claims

The strongest evidence in the cluster is Reuters’ report, which is explicitly attributed to unnamed sources. The Investing.com and Ukrainian news listings appear to relay the same Reuters story rather than provide separate confirmation. No official Alibaba statement, product page or licensing document is included in the supplied evidence.

As a result, the reported plan should not be treated as finalized policy. The sources support the claim that Alibaba is considering a monetization approach for major commercial users. They do not support conclusions about the exact fee, the share Alibaba might seek, the definition of a “big user,” or whether the terms will survive until launch.

There is also no evidence here that customers have agreed to the arrangement or that the next Qwen model has been released. Any claims about performance, adoption or competitive advantage would require additional documentation, such as technical papers, benchmark results, an official license or statements from users.

Implications for builders and enterprises

For startups and product teams, the key planning issue will be license risk rather than model quality alone. Teams considering Qwen for an AI application may need to distinguish between prototyping, internal deployment and revenue-generating use. If Alibaba introduces commercial conditions, a project that begins as a low-cost self-hosted experiment could face new obligations when it reaches scale.

Enterprise buyers will also need to compare the total cost of ownership. Self-hosting an open-source AI model can reduce dependence on a single API provider, but it brings infrastructure, monitoring, security and model-update costs. A revenue-sharing requirement would add another variable to that calculation, particularly for products whose income is directly tied to generated content, automated workflows or AI agents.

For Alibaba, the reported approach could connect model distribution to its wider cloud and enterprise business without abandoning the reach created by an open release. But monetization may also weaken one of the main reasons developers choose an open model: predictable control over deployment and costs. The outcome will depend on whether the final terms are clear, stable and competitive with hosted alternatives.

What to watch next

The first signal will be an official Alibaba announcement describing the next Qwen model, its license and the permitted uses. Builders should look for precise definitions of commercial use, revenue-sharing triggers, geographic limits and whether self-hosted deployments are covered.

Other important signals include the model’s availability through Alibaba Cloud, documentation for local deployment, pricing for inference and comments from early enterprise users. Technical evaluations will also matter: without independent testing, it will be difficult to judge whether the model’s capabilities justify any additional commercial obligations.

Finally, developers should watch whether other model providers adopt similar terms. If large open-model releases begin separating free access from paid commercial scale, licensing and revenue exposure could become as important to model selection as latency, accuracy and infrastructure cost.

Creati.ai perspective

Alibaba’s reported plan highlights the tension at the center of open AI: broad distribution can accelerate adoption, but the cost of training and serving advanced models encourages vendors to reserve monetization rights for the largest businesses. That tension is now moving from theory into licensing and procurement decisions.

For now, the news is a reported plan, not a confirmed pricing policy. AI teams should avoid changing architecture on the basis of the headline alone and wait for Alibaba’s license, deployment rules and commercial thresholds. Those details will determine whether Qwen remains a practical open-source option for businesses—or becomes an open model with a distinctly commercial path for scale users.