Meta is reportedly preparing an AI model that could generate language, code and images, and challenge the position of OpenAI, Microsoft and Google. The plan, described by the Financial Times based on three people familiar with the project, may combine open access with commercial offerings for companies.
An open model with commercial ambitions
The new system may be a version of Meta’s LLaMA, the large language model already used in many open-source projects. Meta has not confirmed that name for the upcoming model; it could be called something else.
The first LLaMA release was made available under a research license, which means it could not be used directly for commercial purposes. Replicas have since appeared. A newer release with open-source terms could give developers and researchers more room to build on Meta’s technology.
Meta CEO Mark Zuckerberg has said the company is working on a new AI model. He has also said Meta wants to use it in its own services and make it available to outside parties. Safety is a particular focus, according to the source article.
Why Meta might invite outside developers
Keeping a model open could help Meta attract a community that improves the technology and creates applications around it. That would let the company’s model spread through projects developed beyond Meta itself, potentially building a larger LLaMA ecosystem.
The approach resembles a strategy in which a company makes a core platform broadly available and gains value as others build on it. The source article compares the possibility to Google’s use of Android. For Meta, a growing ecosystem could help its model compete with services that are more tightly controlled by their creators.
There is also a possible commercial side. The Financial Times report and an insider report from June indicate the model may remain open source, while Meta could retain commercial licensing rights and sell services to businesses. One example mentioned is fine-tuning a model with a company’s own data.
The business model remains uncertain
Meta’s potential offer would rest on a trade-off: developers could use the open model, while companies might pay Meta for licensing or additional support. The appeal to businesses would depend on whether Meta’s complete offering is more useful than relying on open-source access alone.
It is not yet clear which services Meta might charge for, or whether it will charge for fine-tuning. The source article says those details remain unknown. The possible division between open access and paid commercial use is therefore a reported strategy, not a confirmed pricing plan.
If the model is openly available, outside developers could help extend its reach. If Meta also offers services that companies find valuable, the company could benefit from adoption while building a business around the technology. Whether both sides of that approach work depends on the model’s capabilities and on what Meta ultimately makes available.
A challenge to closed AI platforms
An insider familiar with Meta’s high-level strategy told the Financial Times that breaking OpenAI’s dominance is the main goal of the new model. Meta’s head of AI, Yann LeCun, also signaled that open models could reshape competition if they become strong enough to compete with closed systems.
The presentation of the model was described as imminent by Financial Times sources. Until Meta reveals the system and its terms, key questions remain: how capable it will be, how open its release will be, and what businesses may be asked to pay for.
Meta’s reported plan puts openness and commercialization side by side. The company could use community development to spread its technology, then seek business through licensing or services. That combination may give Meta a route into a market currently led by closed platforms, but the outcome will depend on what the company actually launches.