"AI World Android" Llama, no longer popular?
In the first month, Llama 3.1 had only 3.6 million downloads, far below the 5.8 million downloads of the previous version Llama 3. Meta is facing competitive pressure and needs to consider its business strategy. With the increasing download volume of new models from companies like Mistral, Meta may need to give up the commission on cloud service revenue to promote the promotion of Llama by cloud service providers. There are significant differences in business cooperation between open source and closed source models, which poses challenges for the future of Llama
The once highly anticipated Meta open-source large language model, Llama, seems to be losing steam.
According to data from the Hugging Face official website, the latest version of Llama, Llama 3.1, has accumulated approximately 3.6 million downloads from Hugging Face since its release on July 23rd. This is significantly lower than the 5.8 million downloads Llama 3 received in its first month after being released in April.
In contrast, the latest version of Mistral Company's smallest open-source model, Mistral-7B, had more downloads last month than Llama 3.
While Meta's Llama is known for being open-source, competitors such as OpenAI and Anthropic continue to improve their models. Unlike Llama, many developers are willing to pay to use these closed-source options. Despite Llama being free, sometimes using Llama can be more costly than using OpenAI.
Furthermore, the deep integration of Amazon and Microsoft with their competitors has also put pressure on Meta. These two major cloud providers have special agreements with Anthropic and OpenAI, sharing revenue from model customers. They also hold significant equity in these startups and use their own technology in their products.
Open-source and closed-source models have differences in their commercialization models, with open-source models often requiring more flexible business cooperation. Zuckerberg has stated that Meta previously had agreements with cloud service providers to take a portion of the revenue from customers running Llama on cloud servers.
However, now, if Meta wants Llama to be more popular, they may need to make some commercial concessions. By giving up this portion of revenue, cloud service providers can more actively promote Llama to customers.
Zuckerberg is now facing a dilemma. The purpose of Meta developing Llama is to enhance its core business, but the high R&D costs require corresponding returns. If Meta gives up some revenue, it may impact the company's financial situation and subsequently investor confidence.
Zuckerberg also hopes to limit the profit potential of competitors. But to bring Llama to a level comparable to OpenAI's models, continuous investment of significant resources is needed. Balancing this with maintaining the overall financial health of the company requires cost control.
Making reasonable decisions with limited resources is a complex balancing act for Meta