Meta has launched Llama 4 Scout and Llama 4 Maverick, the first models of its Llama herd providing advanced multimodal capabilities.
The models integrate native multimodality, which refers to the ability of an AI model to process and integrate multiple types of data (modalities) such as text, images, audio, and video, within a single framework. The models are inherently designed to handle and combine different forms of data more easily, rather than relying on separate models for each modality.
The firm said the two new models provide better performances on cost across a broad range of widely reported benchmarks than Gemma 3, Gemini 2.0 Flash-Lite, Mistral 3.1, GPT-4o, and comparable results to the new DeepSeek v3 on reasoning and coding and better performances on cost.
Llama 4 Maverick seeks to offer improved performance in image and text understanding, enabling the creation of more sophisticated AI applications.
Llama 4 Scout provides better performance for long-context tasks like multi-document summarisation and reasoning over large codebases, providing understanding both text and images.
“We’re also previewing Llama 4 Behemoth, one of the smartest LLMs in the world and our most powerful yet to serve as a teacher for our new models,”said the company.
The Llama collection aims to provide improved intelligent systems capable of taking generalised actions, conversing naturally with humans, and working through complex problems, with multimodal AI capabilities allowing them to interact more naturally with users, understanding and responding to both visual and auditory cues.
“These models represent the best of Llama, offering multimodal intelligence at a compelling price while outperforming models of significantly larger sizes,” the firm continued.
Llama 4 Scout and Llama 4 Maverick models can be downloaded on Meta’s website, with users able to try Meta AI built with Llama 4 in WhatsApp, Messenger, Instagram Direct.
“As more people continue to use artificial intelligence to enhance their daily lives, it’s important that the leading models and systems are openly available so everyone can build the future of personalised experiences,” Meta added.
Earlier this month, Meta Platforms’ head of artificial intelligence research, Joelle Pineau, announced her departure from the company after eight years, creating a significant vacancy as the tech giant aggressively invests in the AI space.
Pineau, who has served as vice president of AI research and led Meta’s Fundamental AI Research group (FAIR) since 2023, announced her plans to leave in a LinkedIn post on Tuesday. Her last day will be 30 May.
