On April 5, Zuckerberg’s parent company released its next-generation AI models, the Meta Llama 4 Scout and Llama 4 Maverick, signaling another big leap for the giant in on-device efficiency and chatbot performance.
The AI models have been designed to streamline users’ experience across Meta’s platforms Facebook Messenger, Instagram, and WhatsApp, according to the company. Llama 4 Scout, trained on over 10 trillion tokens, is optimized for devices with 16 GB of RAM, while Llama 4 Maverick challenges top-tier chatbots, and Llama 4 Behemoth will be supporting internal model training.
“We are open-sourcing Llama 4 to support innovation in the developer and research community,” Meta said in a statement.
Meta claims Llama 4 Scout outperforms Google’s Gemma 3 and Mistral 3.1, and that Maverick exceeds GPT-4o and Gemini 2.0 Flash on multiple levels. However, according to TechCrunch, Meta’s public version of Maverick may not be similar to the one used for testing, with the giant confirming the benchmark comparisons were based on an “experimental chat version” not publicly available.
Can Meta’s AI Compete?
While Meta Llama 4 appears to rival innovative models in performance, the bigger question is about its long-term effectiveness and reach.
One thing is that the tech giants Meta, Google, in addition to China’s DeepSeek all rely on user data collection to train their AI models. Meta, specifically, owns a spot at the top of this list, with its AI having direct access to billions of social platform information. As much as this is a massive model training bonus for Meta, it has long-term user trust, data ownership, and privacy implications.
The other big question is whether Meta’s Llama 4 will truly compete across platforms or become another restricted AI like xAI’s Grok, which remains locked into X. If Llama 4 stays limited to Meta’s platforms, its ability to scale and challenge the open ecosystem of models like Mistral or Gemma may be undermined.
Meta’s decision to open-source the Llama 4 weights could position it as a leader in transparent AI development but only if its benchmarks are accurate, its models widely accessible, and its data practices transparent.
“The public version of Maverick isn’t the same as the one used in our benchmarks,” a Meta spokesperson highlighted.
Whether Meta Llama 4 is a true competitor or just another platform tool remains unclear, but the knows aspect is that AI wars are heating up, and Meta is determined to be at the center of it.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Intelligent Tech sections to stay informed and up-to-date with our daily articles.