Microsoft 365 Model Comparison Infographic

Using the provided reference image as a benchmark, and employing a concise and easy-to-read business infographic style, create a 16:9 comparison slide to demonstrate how different image generation models render information about Microsoft 365 The existing Japan DX report content has been replaced with four separate sample infographics, arranged in a 2x2 grid on a white background. Layout: Four panels are used, each displaying a different Microsoft 365 ecosystem diagram, with a small horizontal orange model label placed near the left side of each panel. The four labels must be: 1) GPT-Image-2 ,2) GPT-Image-1.5 ,3) MAI Image 2 Efficient and 4) Flux.2 Flex Panel Content: Each panel should center on Microsoft 365, surrounded by application icons and labels. Include recognizable Microsoft-style application tiles/icons in appropriate locations, such as Word, Excel, PowerPoint, Outlook, Teams, OneDrive, SharePoint, OneNote, Forms, Planner, Loop, and Copilot. The top-left panel should be the most refined and have clear text, using a circular radial graph and concise application descriptions. The top-right panel should be simpler, with icons connecting to the central Microsoft 365 box. The bottom-left panel should present a structured category chart, including sections for Productivity, Collaboration, and Storage. The bottom-right panel should present a softer radial graph with category titles and include some imperfections or slight inconsistencies in the labels to highlight the model comparisons. Style: A clean SaaS enterprise infographic aesthetic, ample white space, thin gray connecting lines, Microsoft-style blue/green/orange/purple icons, and clear, legible English text. Retain the practical, business-document feel of the reference images, but do not reproduce their layout, red warning sections, arrows, or original text reminiscent of a Japanese government report. Constraints: Show only four model comparison examples, without additional title banners, watermarks, or Japanese text, and ensure all text is sufficiently clear to compare model performance.