OpenAI innovates with free models, challenging the competition. The tech giant in the field of artificial intelligence positions itself once again at the heart of contemporary issues. With the release of the models gpt-oss-120b and gpt-oss-20b, OpenAI reaffirms its commitment to unprecedented accessibility. Limitless possibilities for research and real-world applications emerging in this new technological landscape. In the face of recurring criticism, the company reconnects with its values of openness and knowledge sharing. This masterstroke comes as the importance of equal access to technologies rapidly advances.
OpenAI unveils new free models
OpenAI recently announced the release of two artificial intelligence models available for free download. These models, named gpt-oss-120b and gpt-oss-20b, introduce an innovative concept, the open-weight language models. This initiative is part of a desire to compete with similar offerings from American and Chinese companies, such as Meta and DeepSeek.
A turning point for the company
Sam Altman, the CEO of OpenAI, emphasized that this approach aims to respond to growing pressure to make the mechanisms of technology more accessible. His statement in a Reddit forum resonated positively, invoking a return to the core values of OpenAI, founded as a non-profit organization in 2015. The goal remains to ensure that Artificial General Intelligence (AGI) benefits all of humanity.
Features of the new models
The models gpt-oss-120b and gpt-oss-20b feature characteristics suited for artificial intelligence tasks. Their performance stands out with a favorable cost-effectiveness ratio, proving capable of conducting internet searches or executing code. OpenAI has designed these models to be easily run on local computer systems, thereby optimizing their accessibility for various users.
Collaboration with partners
OpenAI is actively collaborating with partner companies to explore real-world applications of the new models. Among these partners are major players such as the French telecommunications giant Orange and the cloud data platform Snowflake. These collaborations aim to transform the models into tangible solutions, addressing varied market needs.
Abuse prevention
Regarding the potential misuse of the new models for malicious purposes, OpenAI has integrated mechanisms to reduce the risks of abuse. The company is committed to refining its models with a focus on safety. Altman previously expressed a reflection on the ethics of transparency stating that the company had been on ‘the wrong side of history‘ in terms of communication about its technologies.
A future under committee
Despite tensions around the transition to a profitable entity, OpenAI maintains its status as a non-profit organization while exploring avenues for its revenue-generating branch to operate under the supervision of its board of directors. This choice aims to reconcile investor expectations with OpenAI’s original mission, seeking to ensure that innovation in artificial intelligence occurs ethically.
Frequently asked questions
What are the new models of OpenAI recently launched?
OpenAI has recently launched two new open-weight language models, called gpt-oss-120b and gpt-oss-20b, which can be downloaded for free and modified by users.
Why did OpenAI decide to release these models for free?
The release of these models aligns with OpenAI’s commitment to sharing its expertise and staying true to its initial mission of developing artificial intelligence that benefits all of humanity.
How does an open-weight model function in the context of generative AI?
An open-weight model allows users to access the model’s trained parameters, enabling them to tailor and refine it according to their specific needs.
What types of applications can be created with these new models?
These models are suitable for AI tasks such as internet research and code execution and are designed to be easily run on local computer systems.
What measures has OpenAI taken to prevent abuse of these models?
OpenAI has adjusted the models to make them less likely to be used for malicious purposes, ensuring their responsible use.
How do these new models compare to those of the competition?
OpenAI’s models are designed to compete with similar offerings from American and Chinese companies while differentiating themselves through their open nature and accessibility.
Will OpenAI remain a non-profit organization after these launches?
Yes, OpenAI will continue to operate as a non-profit organization while allowing its revenue-generating branch to function under the supervision of its non-profit board of directors.
Who are OpenAI’s partners for the application of the new models?
OpenAI collaborates with several partners, including the French telecommunications company Orange and the cloud-based data platform Snowflake, to explore concrete use cases for the models.
