Multimodal retrieval and classification across different media types, including text, images, video, and audio, have traditionally relied on dual-encoder models that align visual and textual representations through contrastive learning. The release of Gemini Embedding 2, Google's first natively multimodal embedding model, has raised competition among multimodal retrieval systems. At the same time, frontier Large Language Models (LLMs) have demonstrated strong visual understanding, raising the question of whether they can serve as effective zero-shot rankers. Our study provides the first direct comparison of native multimodal embeddings against LLM-based visual ranking on Flickr30k. We observe that GPT-4.1 and Claude Sonnet 4.6 perform on par with Gemini Embedding 2. Additionally, once embeddings are precomputed, multimodal embeddings are better suited for low-latency applications. Blogger's Review: This article reveals the potential of frontier LLMs in multimodal retrieval tasks and highlights the advantages of native multimodal embeddings in low-latency applications, demonstrating the rapid development of multimodal technology and the broad application prospects of LLMs.