Meta Unveils LlamaCon: First AI Dev Conference

# meta # llamacon

Meta just dropped some big news: they're throwing their first-ever developer conference all about generative AI, and they're calling it LlamaCon. It's happening on April 29, and they're super excited to spill the beans on their latest open-source AI stuff. They want to help developers whip up some cool apps and products, so keep an eye out for more details coming soon. Oh, and don't forget, their usual big bash, Meta Connect, is still on the calendar for September. A few years back, Meta decided to go all in on an "open" approach to AI, hoping to spark a whole ecosystem of apps and platforms. They haven't said exactly how many apps or services are using it, but big names like Goldman Sachs, Nomura Holdings, AT&T, DoorDash, and Accenture are on board with Llama. Plus, they've racked up hundreds of millions of downloads, and they've got at least 25 partners, including heavy hitters like Nvidia, Databricks, Groq, Dell, and Snowflake. Some of these partners have even built extra tools to make Llama work better with proprietary data and run smoother. But here's the twist: Meta got a bit blindsided by DeepSeek, a Chinese AI company that's been making waves with its own "open" AI. Word on the street is that DeepSeek's latest model might just outshine the next Llama version, which is due out soon. Meta's been burning the midnight oil, setting up war rooms to figure out how DeepSeek managed to cut costs on running and deploying models. They're hoping to use those insights to give Llama a boost. Meta's not messing around when it comes to AI—they're planning to drop up to $80 billion on AI projects this year, from hiring AI experts to building new data centers. CEO Mark Zuckerberg's been talking up plans to roll out a bunch of new Llama models over the next few months, including some that can reason like OpenAI's o3-mini and others that can handle multiple types of data right out of the gate. He's also hinted at "agentic" features, meaning future Llama models might be able to do some things on their own. During Meta's Q4 2024 earnings call back in January, Zuckerberg said, "I think this very well could be the year when Llama and open source become the most advanced and widely used AI models. Our goal for Llama this year is to lead." But it's not all smooth sailing for Meta. They're tangled up in a lawsuit over allegedly using copyrighted book materials to train their models without permission. And in the EU, some countries have thrown a wrench in Meta's plans to launch Llama, citing data privacy issues. Some launches have been delayed, and others canceled altogether. *Want to stay in the loop on all things AI? Sign up for TechCrunch's AI-focused newsletter and get it delivered to your inbox every Wednesday.*

Related article

Comments (15)