Artificial intelligence has been in languages for weeks, main thanks to OpenAI employees and its ChatGPT and Bing Chat. Until now, however, the creators of the company with the word “open” in the name jealously guarded their secrets and shared them only with partners. That could change.

Artificial intelligence from OpenAI as open source

According to the latest reports, ChatGPT and GPT-3.5 and GPT-4 supported by API Key will not be the only “export goods” of OpenAI in the segment of artificial intelligence processing text data. According to XDA Developers, its language model can be made available on an open-source basis.

The Information sources report that OpenAI is working on a new model of artificial intelligence. Its source code, in contrast to the company’s existing LLM solutions, would be available to all interested parties.

The new open-source language model is to be developed independently of the closed GPT-3.5 and GPT-4.

OpenAI is not the first company to make its AI available on an open-source basis. A similar step was taken a few months ago by Meta, which is developing its own “large language model”, Meta AI (LLaMA).

It is also an open question how advanced the new OpenAI model will be since the company offers GPT-3.5 and GPT-4 access commercially (as part of the ChatGPT service and through the API for developers and private users). It also makes them available to its partners such as Microsoft.

It is also worth noting that information about at least a partial transition by OpenAI to the open-source model appeared shortly after the Google I/O 2023 conference. During it, the Mountain View company, which is afraid of open-source AI, changed artificial intelligence in all possible cases.