1. Home
  2. / Science and Technology
  3. / Goodbye, OpenAI's ChatGPT? Nvidia launches new AI model that blows away the competition
reading time 5 min read Comments 3 comments

Goodbye, OpenAI's ChatGPT? Nvidia launches new AI model that blows away the competition

Published 19/10/2024 às 18:46
Goodbye, OpenAI's GPT-4? Nvidia launches new AI model that blows away the competition
Photo: NVIDIA

Nvidia surprises by launching a new AI model that surpasses OpenAI's GPT-4 and promises to revolutionize the field of artificial intelligence. Check out the details and understand the impact of this innovation on the competition!

Nvidia has made a major move into artificial intelligence by launching a new AI model that has outperformed offerings from major companies in the sector, such as OpenAI and Anthropic. Although the launch was done quietly, the results speak for themselves, positioning Nvidia as a powerhouse not only in hardware, but also in AI software.

A discreet but impactful launch

Last Tuesday, Nvidia made available the model called Llama-3.1-Nemotron-70B-Instruct on the AI-focused Hugging Face platform, without much fanfare.

The model quickly gained attention for its performance in major benchmark tests. It achieved 85,0 on Arena Hard, 57,6 on AlpacaEval 2 LC, and 8,98 on GPT-4-Turbo MT-Bench, surpassing the scores of renowned models such as OpenAI’s GPT-4 and Anthropic’s Claude 3.5 Sonnet.

This achievement represents an important milestone for Nvidia, a company traditionally known for its GPUs (processing units) chart), essential for training large AI models. However, by launching its own language model, the company shows its ability to compete directly with software giants.

The top of the Chatbot Arena ranking. Source: LLMArena

The technology behind the Llama-3.1-Nemotron-70B-Instruct

Nvidia used Meta’s open-source Llama 3.1 model as the basis for developing Nemotron. The process included advanced training techniques such as Reinforcement Learning from Human Feedback (RLHF), which allows the AI ​​to adjust its responses based on human preferences.

This means that the model can offer more natural and contextualized responses, which is a big differentiator compared to competing models.

A practical example of this ability was demonstrated in a simple test, where the model correctly answered the question “How many 'r's are there in the word 'strawberry'?”, demonstrating a high level of understanding and clarity in its responses.

The concept of “alignment” is one of the main advantages of Nvidia’s new model. This term refers to the ability of AI to generate responses that match the needs and preferences of its users.

In practice, this means fewer errors and greater customer satisfaction, which is crucial for companies that rely on AI to improve customer service and automate processes.

Implications for the business market

For companies looking for AI solutions, the Llama-3.1-Nemotron-70B-Instruct offers a robust and affordable alternative.

Nvidia makes the model freely available for inference through its build.nvidia.com platform, making it easier for companies of all sizes to access cutting-edge AI.

Personalization is another strength of the model. Many companies need AI that can be tailored to specific tasks, such as customer service or detailed reporting.

Nvidia's model allows for this flexibility, making it a valuable tool for industries ranging from finance to healthcare.

However, Nvidia warns that the model has not yet been fully tuned to handle areas that require maximum precision, such as advanced mathematics or legal reasoning.

This means that companies will need to implement additional security measures to ensure proper use of AI, minimizing the risk of errors.

A new chapter in the AI ​​arms race

The launch of Llama-3.1-Nemotron-70B-Instruct signals a shift in the competition for advanced AI models. Until now, companies like OpenAI and Anthropic have dominated the development of large language models, but Nvidia has shown that it can be a strong competitor as well. By expanding its focus from hardware to software, Nvidia is putting pressure on its rivals to accelerate their own innovations.

Nvidia’s strategic shift is also reflected in the recent launch of the NVLM 1.0 family of models, including the 72-billion-parameter NVLM-D-72B, another significant advancement that solidifies the company’s position as a leader in AI. These multimodal models can interpret and process not only text, but also images, expanding the range of possible applications.

The Future of AI and Nvidia's Role

As Llama-3.1-Nemotron-70B-Instruct is tested and deployed across a range of industries, new applications are expected to emerge. Companies in sectors such as healthcare, education and finance are already exploring how the model can be integrated into their systems to automate processes and improve efficiency.

However, the long-term success of this model will depend on its ability to translate its impressive benchmark scores into practical solutions. The AI ​​community will be watching closely to see how the model performs in real-world scenarios, beyond controlled testing environments.

If Nvidia can continue to innovate and expand its AI offerings, we’re likely to see a shake-up of the industry in the coming years. The company has already demonstrated that it has the ability to compete with giants like OpenAI, but the real test will be the widespread adoption of its solutions.

The launch of Llama-3.1-Nemotron-70B-Instruct marks a turning point in the race for AI leadership. Nvidia has not only proven that it can compete in AI software development, but that it is willing to challenge the status quo.

Companies across all industries now have at their disposal a powerful, affordable and flexible new tool that can be tailored to their specific needs.

Register
Notify
guest
3 Comments
Older
Last Most voted
Feedbacks
View all comments
Fabio Lucas Carvalho

Journalist specializing in a wide range of topics, such as technology, politics, shipbuilding, geopolitics, renewable energy and economics. I have been working since 2015 with prominent publications in major news portals. My degree in Information Technology Management from Faculdade de Petrolina (Facape) adds a unique technical perspective to my analyses and reports. With over 10 thousand articles published in renowned media outlets, I always seek to bring detailed information and relevant insights to the reader. For story suggestions or any questions, please contact me by email at flclucas@hotmail.com.

Share across apps
0
We would love your opinion on this subject, comment!x
()
x