Meta Llama Model Downloads Surpass One Billion, Showing Explosive Growth!

AIbase基地

Published inAI News · 4 min read · Mar 19, 2025

Meta's family of open-source AI models, Llama, recently reached a significant milestone: over one billion downloads. This news was announced by CEO Mark Zuckerberg on the social media platform Threads. Compared to the 650 million downloads in December 2024, Llama's downloads surged by 153% in just three months.

Meta, Metaverse, Facebook

Initially launched for researchers, Llama has gradually gained traction among global developers and AI enthusiasts. Ahmad Al-Dahle, VP of Meta's generative AI division, shared this achievement on LinkedIn, stating that Llama evolved from a research project to the most widely used open-source AI model in the past two years.

Al-Dahle expressed pride in his team's efforts, mentioning the two-year anniversary of Meta's generative AI organization on February 27th. He felt both exhilarated and humbled by the team's dedication to global developers, startups, and businesses. He believes Llama's download numbers represent trust in open-source AI and a shared belief in building a better future.

To further propel Llama's development, Meta plans to release several Llama models in the coming months. These include a "reasoning" model similar to OpenAI's o3-mini, and versions with multi-modal capabilities. Zuckerberg also hinted at models with "agent" capabilities, suggesting some models might possess autonomous operational abilities.

Last October, Zuckerberg confirmed during the company's Q3 earnings call that Llama4 would be released in early 2025. He anticipates smaller Llama4 models will be ready first, considering it a significant advancement involving new modalities, enhanced capabilities, stronger reasoning, and faster processing speeds.

While the Llama3 model marked a significant industry turning point, Zuckerberg expressed even greater enthusiasm for Llama4's development. He mentioned that Llama4's training involved over 100,000 H100 chips, far exceeding the scale known to other companies.

Llama already supports Meta's platforms like Facebook, Instagram, and WhatsApp, and plays a crucial role in the company's long-term AI strategy.

Key Highlights:
📈 Llama downloads surpass 1 billion, a 153% increase in three months.
👩‍💻 Meta plans to release various new models, including reasoning and multi-modal capabilities.
🚀 Zuckerberg is confident about Llama4's development, anticipating enhanced performance.

Challenging Conventions: A Breakthrough Transformer Architecture Without Normalization Layers

In the field of deep learning, normalization layers are considered an indispensable component of modern neural networks. Recently, research led by Meta FAIR research scientist Zhuang Liu, titled "Transformer without Normalization Layers", has garnered significant attention. This research not only introduces a new technique called Dynamic Tanh (DyT), but also demonstrates the effectiveness of Transformer architectures even without traditional normalization layers.

Anthropic, IBM, and Meta Tech Leaders Warn AI to Replace Software Developers

At a recent international conference, Anthropic CEO Dario Amodei made a striking prediction that AI will take over 90% of code writing within the next three to six months. Amodei suggested that if this trend continues, AI could almost entirely replace human programmers within 12 months. He noted that while programmers may still need to set specific parameters and goals for AI, this process too could potentially be automated by technology in the future.

CMU Team Introduces Meta Reinforcement Fine-Tuning: A Novel Paradigm for Enhancing Large Language Model Reasoning

Large Language Models (LLMs) are constantly evolving in the field of artificial intelligence. Researchers from Carnegie Mellon University (CMU) and HuggingFace recently introduced a new method called Meta Reinforcement Fine-Tuning (MRT). This method aims to optimize the computational efficiency of LLMs during testing, particularly excelling in solving complex reasoning problems. Studies show that existing LLMs struggle with...

Meta Aims to Slash $65 Billion AI Bill with Custom Chips, Seeking to Break Free from Nvidia?

Social media giant Meta is facing unprecedented AI infrastructure costs, with projected AI-related spending soaring to a staggering $65 billion, potentially contributing to a total annual expenditure of $119 billion! Faced with this astronomical bill, the tech giant has decided to develop its own AI chips, and has already made significant progress. Recent reports indicate that Meta is about to begin a small-scale deployment of its custom chips, signaling a gradual move away from Nvidia and its reliance on their GPUs.