Meta's LlamaCon Focuses on Competing with OpenAI

On Tuesday, Meta introduced its inaugural AI developer conference, LlamaCon, at its Menlo Park, California headquarters. The tech giant unveiled a consumer-facing Meta AI chatbot app designed to compete with ChatGPT, along with a developer API for accessing its Llama models via the cloud.

Both launches aim to broaden adoption of Meta’s open Llama AI models, but this expansion also serves another key goal: challenging OpenAI. In essence, Meta seeks to cultivate a vibrant open AI ecosystem that stands in contrast to “closed” providers like OpenAI, which restrict access behind proprietary services.

Meta’s new AI chatbot app appears to preempt OpenAI’s rumored social networking ambitions. It features a social feed where users can share their AI conversations and delivers personalized responses based on individual activity across Meta’s apps.

As for the Llama API, it directly competes with OpenAI’s API services. Designed for streamlined developer use, the Llama API enables app integration with cloud-hosted Llama models using just one line of code. By removing the reliance on third-party cloud providers, Meta can offer AI developers a more complete toolkit.

Meta, like numerous AI firms, considers OpenAI a primary competitor. Internal documents from a legal case show Meta executives were intensely focused on surpassing OpenAI’s GPT-4, once the industry’s leading model. Undermining proprietary AI providers like OpenAI has long been central to Meta’s AI playbook. In a July 2024 letter, Meta CEO Mark Zuckerberg emphasized the contrast with companies such as OpenAI, stating that “selling access to AI models isn’t [Meta’s] business model.”

Several AI researchers who spoke with TechCrunch ahead of LlamaCon were anticipating Meta’s release of a competitive reasoning model similar to OpenAI’s o3-mini. That release did not occur. For Meta, however, leading the AI race is not the only measure of success.

During an onstage discussion with Databricks CEO Ali Ghodsi at LlamaCon, Zuckerberg expressed that any AI lab offering open models—including DeepSeek and Alibaba’s Qwen—are partners in opposing closed-model providers.

“A key benefit of open source is the ability to mix and match. If another model, such as DeepSeek, excels in one area—or if Qwen is better at something else—developers can combine the best aspects of different models to create exactly what they need,” Zuckerberg explained. “This is how I believe open source will eventually surpass the quality of all closed-source models… It feels like an unstoppable momentum.”

Beyond curbing OpenAI’s expansion, Meta’s open-model push may also help it qualify for regulatory benefits. The EU AI Act provides specific exemptions for companies that distribute “free and open source” AI systems. Although debate continues over whether Llama models meet all open-source criteria, Meta consistently describes them as “open source.”

Whatever the motivation, Meta appears committed to advancing AI innovations that reinforce the open-model ecosystem and restrict OpenAI’s influence—even if it means sometimes lagging in delivering the most advanced models itself.