NVIDIA New NVLM vs GPT-4o, Claude, Gemini

Author(s): Lo Zarantonello

Originally published on Towards AI.

It actually outperforms GPT-4o, Claude 3.5 Sonnet, and Gemini 1.5 Pro in some categories!

This member-only story is on us. Upgrade to access all of Medium.

In the last few months, the new frontiers LLMs race intensified with a new entrant: NVIDIA.

nvidia/NVLM-D-72B is available on HuggingFace and Megatron

NVIDIA has achieved the status of one of the most valued companies in the world. Now, it moves into creating open-source frontier models that rival the best-in-class open and close models.

The new “NVLM 1.0, a family of frontier-class multimodal large language models (LLMs) that achieve state-of-the-art results on vision-language tasks, rivaling the leading proprietary models (e.g., GPT-4o) and open-access models” — HuggingFace

The model is currently available on

HuggingFaceMegatron

There are at least two major reasons why NVLM 1.0 is important:

Business: It opens a new line of business for NVIDIA and introduces another player in the AI oligopoly. Sam didn’t say “Do not invest in NVIDIA”, right?Also, will Intel follow or focus on restructuring?Technical: It seems to be different than other models!

Well, every model claims so!

Let’s see what they say!

NVIDIA

According to NVLM: Open Frontier-Class Multimodal LLMs, NVLM 1.0 can understand the meme below.

In particular, the “model can understand the humor behind the “abstract vs. paper” meme in example (a) by performing OCR to recognize the text labels for each image and… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI