OpenAI on Monday launched GPT-4.1, its latest family of generative AI models.
The release includes three models: GPT-4.1, a smaller GPT-4.1 mini and GPT-4.1 nano. On a livestream announcement, OpenAI chief product officer Kevin Weil called GPT-4.1 nano the developer’s «smallest, fastest and cheapest model ever.»
Developers will be able to access the models through OpenAI’s API.
The new family of models will only be available through the API for developers. OpenAI said many improvements in the latest models have been incorporated into GPT-4o for ChatGPT.
As the AI race continues, OpenAI has repeatedly made headlines for its ability to introduce more powerful models to an increasing number of users. Last month, OpenAI introduced image generation in ChatGPT-4o, which impressed with its accuracy and realism. The feature became so popular that OpenAI had to set rate limits after CEO Sam Altman posted, «Our GPUs are melting.» Last week, OpenAI increased the memory capabilities of ChatGPT so that it could do a better job of recalling information from past conversations. This comes as Google, too, is pushing to lead the AI space with more offerings, including a powerful and efficient new AI chip called Ironwood. The company launched a «practically unlimited» coding assistant for free earlier this year and is currently rolling out Veo 2, a new generator tool to compete with OpenAI’s Sora.
While OpenAI isn’t profitable and is trying to restructure itself from a nonprofit to a for-profit organization, it’s currently valued at $300 billion. Even though ChatGPT is still far from overtaking Google, it was the most downloaded app on app stores last month.
CNET’s Jon Reed contributed to this report.
