OpenAI is taking on Mark Zuckerberg’s Meta by launching its own freely available artificial intelligence models.
The ChatGPT developer has announced two open-weight large language models, which are free to download and can be customised by a user.
Meta’s Llama models are available on a similar basis, and OpenAI’s move marks a departure from ChatGPT, which is based on a “closed” model that cannot be customised.
Sam Altman, OpenAI’s chief executive, said the company was excited to add to a stack of freely available AI models “based on democratic values … and for wide benefit”.
He added: “We’re excited to make this model, the result of billions of dollars of research, available to the world to get AI into the hands of the most people possible.”
Zuckerberg, similarly, has said that making models freely available and customisable will “ensure that more people around the world have access to the benefits and opportunities of AI” and that power is not concentrated in the hands of a small number of companies.
However, Meta has said his company may need to be “careful” about taking a similar approach with highly advanced AI models.
DeepSeek, a Chinese rival to OpenAI and Meta, has also released powerful models that can be downloaded and customised freely.
OpenAI said the two models, called gpt-oss-120b and gpt-oss-20b-two, outperformed similarly sized models on reasoning tasks, with the larger, 120b model achieving a near-equal performance to its o4-mini model in terms of core reasoning.
Open AI added that during the testing process it had created “maliciously fine-tuned” versions of the models to simulate biological and cybersecurity threats but found they were “unable to reach high capability levels”.
The existence of freely available, powerful AI models that can be customised by developers has alarmed some experts who have warned that they could be adapted to cause serious harm, such as developing bioweapons.
Meta has defined its Llama models as “open source”, meaning they can also be downloaded and customised freely, while their training datasets, architecture and training code are also provided.
However, the Open Source Initiative, a US-based industry organisation, has said the restrictions Meta places on using its models mean it does not qualify as fully open source. OpenAI has used the term “open weight”, effectively a notch down from open source, which means the model can still be fine tuned – but its workings are not fully transparent to the developer.
OpenAI’s announcement came amid speculation that a new version of the model that underpins ChatGPT could be released soon. On Sunday Altman shared a screenshot of what appeared to be the company’s latest AI model, GPT-5.
Meanwhile, Google has outlined its latest step towards artificial general intelligence (AGI) with a new model that allows AI systems to interact with a convincing simulation of the real world.
The Genie 3 “world model” could be used to train robots and autonomous vehicles as they engage with realistic recreations of environments such as warehouses, according to Google.
The US technology company’s AI division, Google DeepMind, argues that world models are a key step to achieving AGI, a hypothetical level of AI where a system can carry out most tasks on a par with humans – rather than just individual tasks such as playing chess or translating languages – and potentially do someone’s job.
DeepMind said such models would play an important role in the development of AI agents, or systems that carry out tasks autonomously.
“We expect this technology to play a critical role as we push toward AGI, and agents play a greater role in the world,” DeepMind said.