On Tuesday, Meta announced that it will hold its first development meeting dedicated to generating AI.
After the Rama family of meta-generating AI models, it is called the Ramacon and the meeting is scheduled to take place on April 29th. Meta said it would be sharing the latest (its) open source AI development. Products.”
According to Meta, additional details will be available soon. The company’s annual developer conference, Meta Connect, will be held in the second half of September. This is a typical window.
A few years ago, Meta embraced a “open” approach to developing AI technologies to promote an ecosystem of apps and platforms. It has not been revealed how many apps and services are built on top of it, but it has been pointed out previously that Goldman Sachs, Nomura Holdings, AT&T, Doordash and Accenture will use Llama. The company claims it has hundreds of millions of downloads of the model and has at least 25 partners hosting Llama, including Nvidia, Databricks, Groq, Dell and Snowflake. Make your own data and allow them to run with low latency.
However, Meta reportedly fell thinly due to the rise of Chinese AI company Deepseek.
Meta reportedly believes that one of Deepseek’s new models could be better than the next version of Llama, which is due to be released in the next few weeks. Meta is said to have scrambled Deepseek to set up a war room to reduce the costs of running and deploying the models, so those learnings can be applied to the development of llamas themselves.
Meta recently said it would spend up to $80 billion on AI-related projects this year, including AI employment and the construction of new AI data centers.
Meta CEO Mark Zuckerberg previously announced several Llama models over the coming months, including “inference” models along the line of Openai’s O3-Mini and “inference” models along the model with native multimodal capabilities. It has been announced that it is planning to launch a model. He also suggests a “agent” function, suggesting that future llama models can autonomously carry out certain actions.
“I think this is the year when llamas and open source become the most advanced and widely used AI model,” Zuckerberg said in a revenue call for the fourth quarter 2024 in the January Meta. . “(o) (this year’s llama) goal is to lead.”
Meta is also in the midst of a lawsuit accusing them of training the model on this copyrighted material. Another challenge to Meta’s Lama’s ambitions has forced several EU countries to postpone the company.
TechCrunch has a newsletter focused on AI! Sign up here to get it every Wednesday in your inbox.