NVLM 1.0: Nvidia launches open AI model to rival OpenAI

Nvidia announced on Wednesday (2) the arrival of a new open source LLM (Large Language Model) to compete with GPT-4o. Called NVLM 1.0, this family of multi-mode models focuses on versatility and ease of exploring developers’ capabilities.

The material was published by the company in an article in which the 72 billion parameters available in the NVLM-D-72B model were highlighted. The technology, which focuses entirely on artificial intelligence, can interpret images and produce correct answers about the content, but for now it can only provide results in text format.

In one example Nvidia uses, the brand uses a kitten meme and asks the model to explain why the image is funny. In short, LLM was able to interpret and explain the meme in a comfortable and entertaining way, revealing the power of Nvidia’s new version.

The model may serve as an industry standard

The fact that a great language model like NVLM 1.0 comes in an open format is good news for developers and researchers in this field. Unlike other existing models that have a proprietary structure, launching a green team can help smaller and independent companies develop their infrastructure.

To explain the new family’s success, Nvidia explains that it “built and integrated a high-quality, text-only dataset with a significant amount of multimodal math and reasoning data into multimodal training, resulting in improved math and coding capabilities.” ”.

A strong open source model could compel other major tech companies to follow suit. Considering the hardware and software power that Nvidia has used with the rise of artificial intelligence in recent years, it is thought that the NVLM 1.0 family can set a new standard in the industry.