Meta’s latest AI model, Llama 4, is here: Can it generate Ghibli art

Meta has launched the first two models in its Llama 4 series—Llama 4 Scout and Llama 4 Maverick—ushering in its next-generation open-weight, multimodal AI models. Both support text and image inputs and feature a 10 million token context window. Llama 4 Scout has 17B active parameters and fits on a single NVIDIA H100 GPU, while Llama 4 Maverick offers enhanced performance with 400B total parameters. The AI language model is now integrated into Meta’s own Meta AI chatbot. While the addition brings improved features and capabilities, the big question is whether it can create Ghibli art like ChatGPT, Grok or other AI chatbots? We tried, and here’s what we’ve found.

The first answer is that yes, Llama 4 has the capabilities to create images and even different art styles, including Ghibli style images. We asked it to create an image showing “a girl waiting at the bus stand”. It did generate the images – 4 of them – at one. It’s very fast. Speed is much appreciated here, but the quality of the output isn’t as good as ChatGPT’s output.

Llama 4 vs ChatGPT

While both ChatGPT and Grok have integrated features that allow users to directly upload and transform their personal photos into the distinctive Ghibli art style, Meta AI currently lacks this specific functionality. Consequently, users of Meta AI are limited to utilizing text-based command prompts to generate Ghibli-inspired artwork. This discrepancy in features highlights the varying capabilities and user experiences across different AI platforms, particularly concerning the generation of artistic content and the integration of user-provided images.

One advantage that Meta AI has right now over the other Chatbots is that it can animate the images that it has created. So, once the images are generated, you can hit the "Animate" button to get the animated Ghibli images.