OpenAI has released a new AI model, named GPT-OSS, that comes in two versions. One will run on Nvidia’s high-end GPUs with 80GB of RAM, and the smaller model, can run on smaller devices with as little as 16GB of memory.
OpenAI’s press release says that these are “designed to be used within agentic workflows with exceptional instruction following, tool use like web search or Python code execution, and reasoning capabilitiesāincluding the ability to adjust the reasoning effort for tasks that donāt require complex reasoning and/or target very low latency final outputs.”
Both models are out now; there are links to use them on OpenAI’s website. The company seems to be gearing these towards third parties that can then use them in their tools since they are under the Apache 2.0 license. This license allows anyone to use and modify the software as they see fit.
This is the company’s first open model in six years. CEO Sam Altman cited safety concerns with releasing such powerful AI models to the general public without real guardrails. However, according to a report on The Verge, he recounted that stance in January of this year after seeing how popular and how much change open-source models from other companies have created.
The report also mentions that the company noticed that its customers were already using open-source models for various tasks, so it figured it might as well release a new model so that those customers could use OpenAI tech across the board.
That said, according to OpenAI, the model is still quite safe, and various safety firms have tested it to ensure that it won’t pose any major cybersecurity or regular safety risks once it hits the mass market.
Source: OpenAI Via: The VergeĀ
MobileSyrup may earn a commission from purchases made via our links, which helps fund the journalism we provide free on our website. These links do not influence our editorial content. Support us here.
