OpenAI plans to release an open-source language model that surpasses its peers

OpenAI plans to release its first open-source language model since GPT – 2 in early summer this year, aiming to surpass the capabilities of similar reasoning models. Currently, this model is in the early development stage and is led by Aidan Clark, the Vice President of Research at OpenAI.

OpenAI hopes to avoid the criticism that previous open-source models like Llama and Google’s Gemma faced due to excessive restrictions by adopting a permissive license agreement that allows for nearly unrestricted use and commercialization.

This new model will follow an “input text, output text” mode. Developers may be given the freedom to choose whether to enable or disable the reasoning function. Although enabling the reasoning function can improve accuracy, it will also introduce latency.

If the response is positive, OpenAI plans to continue releasing more models, including smaller versions.