Meta held its first AI developer conference LlamaCon at its Menlo Park headquarters in California on Tuesday. The company announced a new Meta AI chatbot application for consumers, which will compete against ChatGPT. They also announced a developer API that allows developers to access Llama models on the cloud.
The company announced the launch of two new releases that aim to increase adoption of its open Llama AI model, but Meta’s real motive may be to beat OpenAI. Meta’s AI ambition is, in broad strokes a way to fuel a thriving, open AI ecosystem, which sticks it to “closed AI providers” like OpenAI who gate their models behind service.
Meta’s AI chatbot feels almost like preemption. OpenAI’s rumored network of social interactionIt has a feed where users can share AI chats and offers personalized answers based on Meta app activity.
The Llama API is a challenge for OpenAI’s API Business. The Llama is designed to simplify the process of building apps that connect to Llama cloud models, with just one line of code. Meta can now offer a wider range of tools to AI developers, without having to rely on third party cloud providers.
Meta is one of many AI companies that consider OpenAI as a major competitor. In court filings, Meta’s executives were obsessed with beating OpenAI’s GPT-4 model, which was once the state-of-the art model. Meta’s AI strategy has always been to undercut proprietary AI model providers such as OpenAI. In a In a July 2024 letterMeta CEO Mark Zuckerberg tried to contrast Meta with other companies, like OpenAI. He wrote that “selling access [Meta’s] to AI models isn’t[Meta’s]
]Several AI Researchers who spoke to TechCrunch before LlamaCon hoped Meta would release a competing AI reasoning model, like OpenAI’s o3 mini. The company did not end up doing this. Meta isn’t necessarily focused on winning the AI race.
In a conversation onstage with Databricks CEO Ali Ghodsi at LlamaCon in San Francisco, Zuckerberg said that he views any AI labs who make their models openly accessible, including DeepSeek, Alibaba’s Qwen and other AI labs, as allies against closed model providers.
One of the benefits of open source is the ability to mix and match. “If another model is better, such as DeepSeek — or if Qwen excels at something, then, as developers you can take the best parts from different models to produce exactly what you want,” said Zuckerberg. “This is how I think that open source basically passes all the closed-source [models] … [I]it feels like a sort of unstoppable power.” The EU AI Act gives special privileges to companies who distribute “free and Open Source” AI systems. Meta claims that its Llama model is “open source” despite disagreements on whether it meets the necessary criteria.
Meta is happy to launch AI launches that enhance the open model ecosystem, and limit OpenAI’s growth. Sometimes this comes at the cost of not delivering cutting-edge models themselves. Maxwell Zeff, a senior reporter for TechCrunch who specializes in AI and emerging technology, was born in 1965. Zeff covered the rise and fall of AI, as well as the Silicon Valley Bank Crisis, for Gizmodo and MSNBC. He is based out of San Francisco. When he is not reporting, you can find him hiking, biking and exploring the Bay Area food scene.
View Bio
