{"id":93067,"date":"2025-04-06T02:31:39","date_gmt":"2025-04-06T02:31:39","guid":{"rendered":"https:\/\/neclink.com\/index.php\/2025\/04\/06\/meta-introduces-llama-4-with-two-new-ai-models-available-now-and-two-more-on-the-way\/"},"modified":"2025-04-06T02:31:39","modified_gmt":"2025-04-06T02:31:39","slug":"meta-introduces-llama-4-with-two-new-ai-models-available-now-and-two-more-on-the-way","status":"publish","type":"post","link":"https:\/\/neclink.com\/index.php\/2025\/04\/06\/meta-introduces-llama-4-with-two-new-ai-models-available-now-and-two-more-on-the-way\/","title":{"rendered":"Meta introduces Llama 4 with two new AI models available now, and two more on the way"},"content":{"rendered":"


\n<\/p>\n

Meta has released the first two models from its multimodal Llama 4 suite: LLama 4 Scout and Llama 4 Maverick. Maverick is \u201cthe workhorse\u201d of the two and excels at image and text understanding for \u201cgeneral assistant and chat use cases,\u201d the company said in a blog post<\/a>, while the smaller model Scout could tackle things like \u201cmulti-document summarization, parsing extensive user activity for personalized tasks, and reasoning over vast codebases.\u201d The company also introduced Llama 4 Behemoth, an upcoming model it says is \u201camong the world\u2019s smartest LLMs\u201d \u2014 and CEO Mark Zuckerberg<\/a> said we\u2019ll be hearing about a fourth model, LLama 4 Reasoning, \u201cin the next month.\u201d <\/p>\n

Both Maverick and Scout are available to download now from the LLama<\/a> website and Hugging Face<\/a>, and they\u2019ve been added to Meta AI, including for WhatsApp, Messenger and Instagram DMs. <\/p>\n

<\/span><\/p>\n

\"A
<\/figcaption>
Meta<\/div>\n<\/figure>\n

Scout has 17 billion active parameters with 16 experts, Meta says. According to Zuckerberg, \u201cIt\u2019s extremely fast, natively multimodal, and has an industry leading, nearly infinite 10 million token context length, and it is designed to run on a single GPU.\u201d Maverick on the other hand has 17 billion active parameters with 128 experts. The company says it beats competitors like GPT-4o and Gemini 2.0 on coding, reasoning, multilingual, long-context and image benchmarks, and stacks up against DeepSeek v3.1 on reasoning and coding.<\/p>\n

Zuckerberg is already calling the upcoming Behemoth model, which is still training, \u201cthe highest performing base model in the world,\u201d with 288 billion active parameters, according to the company. It may not be here yet, but it\u2019s likely we\u2019ll be hearing a lot more about that and the Reasoning model soon; Meta\u2019s big AI developer conference, LlamaCon<\/a>, is just a few weeks away.<\/p>\n

This article originally appeared on Engadget at https:\/\/www.engadget.com\/ai\/meta-introduces-llama-4-with-two-new-models-available-now-and-two-more-on-the-way-214524295.html?src=rss