English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The surge of interest towards Multi-modal Large Language Models (MLLMs),e.g., GPT-4V(ision) from OpenAI, has marked a significant trend in bothacademia and industry. They endow Large Language Models (LLMs) with powerfulcapabilities in visual understanding, enabling them to tackle diversemulti-modal tasks. Very recently, Google released Gemini, its newest and mostcapable MLLM built from the ground up for multi-modality. In light of thesuperior reasoning capabilities, can Gemini challenge GPT-4V's leading positionin multi-modal learning? In this paper, we present a preliminary exploration ofGemini Pro's visual understanding proficiency, which comprehensively coversfour domains: fundamental perception, advanced cognition, challenging visiontasks, and various expert capacities. We compare Gemini Pro with thestate-of-the-art GPT-4V to evaluate its upper limits, along with the latestopen-sourced MLLM, Sphinx, which reveals the gap between manual efforts andblack-box systems. The qualitative samples indicate that, while GPT-4V andGemini showcase different answering styles and preferences, they can exhibitcomparable visual reasoning capabilities, and Sphinx still trails behind themconcerning domain generalizability. Specifically, GPT-4V tends to elaboratedetailed explanations and intermediate steps, and Gemini prefers to output adirect and concise answer. The quantitative evaluation on the popular MMEbenchmark also demonstrates the potential of Gemini to be a strong challengerto GPT-4V. Our early investigation of Gemini also observes some common issuesof MLLMs, indicating that there still remains a considerable distance towardsartificial general intelligence. Our project for tracking the progress of MLLMis released athttps://github.com/BradyFU/Awesome-Multimodal-Large-Language-Models.

A Challenger to GPT-4V? Early Explorations of Gemini in Visual Expertise