openai/gpt-chat-latest → A (4:1)

Side A delivers substantial functional improvements: it adds a fullscreen/chromeless vote-compare layout, updates the POST success path to morph in the newly created vote preview, reworks edge-history display to normalize ratios by page order and sort by vote strength, and includes browser/unit test updates to verify the behavior. Side B is a focused visual refinement that changes rank-row coloring from list-position to score-based min–max normalization with good tests, but it affects presentation only and has a narrower long-term impact.