chat-ui

fda90d7c - Show streaming reasoning in a fixed-height fade viewport (#2262)

Show streaming reasoning in a fixed-height fade viewport (#2262) * Show streaming reasoning in a fixed-height fade viewport While reasoning is streaming, render the <think> content in a bottom-aligned 4rem viewport with a top gradient mask so only the most recent lines stay visible. This avoids the overwhelming wall-of-text effect during long chain-of-thought, and works uniformly across model formats (DeepSeek R1, gpt-oss, Qwen3, etc.) without per-model parsing. After streaming completes, the existing collapsed/expand-to-full behavior is unchanged. * Bump streaming reasoning viewport to 320px * Cap reasoning fade band at ~2 lines (48px) * Invert Thinking header shimmer in light mode * Anchor Thinking shimmer to text-gray-500 baseline * Apply reasoning fade only when content overflows viewport * Vertically center alternatives nav and copy/edit buttons --------- Co-authored-by: Claude <noreply@anthropic.com>