Stage 376: PR #2450

# Conflicts:
#	CHANGELOG.md

CHANGELOG.md

@@ -9,6 +9,7 @@
 - Compact live Thinking cards now reuse the same timeline card across sequential tool calls, preventing repeated Thinking cards from stacking during one multi-tool turn.
 - Refresh context-window metadata when a session's resolved model changes during load or when the user switches models, so high-context models do not stay stuck on a stale prior window and trigger premature compression. Fixes #2442.
 - **PR #2445** by @Michaelyklam (fixes #2443) — `/api/models` now fingerprints model-catalog inputs as part of its persisted cache metadata, so server-side catalog additions and Codex local catalog changes invalidate `models_cache.json` immediately instead of waiting for the 24-hour TTL or manual cache deletion.
+- **PR #2450** by @Michaelyklam (fixes #2447) — Cap the optional streaming word-fade drain after the final `done` SSE event so very large or bursty completed responses are rendered from the canonical session promptly instead of keeping the chat in a live/working state until Stop is pressed.
 
 ## [v0.51.82] — 2026-05-17 — Release BF (stage-375 — 2-PR batch — table renderer pipe protection + Catppuccin appearance skin)
 

static/messages.js

@@ -697,6 +697,7 @@ function attachLiveStream(activeSid, streamId, uploaded=[], options={}){
   const _STREAM_FADE_MAX_MS=350;
   const _STREAM_FADE_STAGGER_MS=16;
   const _STREAM_FADE_DONE_MAX_MS=320;
+  const _STREAM_FADE_DONE_DRAIN_MAX_MS=900;
   const _streamFadeEnabledForStream=window._fadeTextEffect===true;
 
   // rAF-throttled rendering: buffer tokens, render at most once per frame
@@ -1086,6 +1087,8 @@ function attachLiveStream(activeSid, streamId, uploaded=[], options={}){
       : _stripXmlToolCalls(assistantText.slice(segmentStart));
   }
   function _drainStreamFadeBeforeDone(onDone){
+    const drainStartedAt=performance.now();
+    let forcedDone=false;
     const step=()=>{
       if(!assistantBody){onDone();return;}
       const target=_streamFadeCurrentDisplayText();
@@ -1101,6 +1104,15 @@ function attachLiveStream(activeSid, streamId, uploaded=[], options={}){
         setTimeout(onDone, Math.min(remainingAnimationMs, _STREAM_FADE_DONE_MAX_MS));
         return;
       }
+      // Final SSE `done` means the canonical completed session is available.
+      // The optional word-fade playout must not keep that completed answer
+      // hidden behind the live Thinking state for large/bursty responses.
+      if(!forcedDone&&performance.now()-drainStartedAt>=_STREAM_FADE_DONE_DRAIN_MAX_MS){
+        forcedDone=true;
+        if(_smdParser) _smdEndParser();
+        onDone();
+        return;
+      }
       setTimeout(()=>requestAnimationFrame(step), 33);
     };
     step();

tests/test_smooth_text_fade.py

@@ -82,6 +82,7 @@ const _STREAM_FADE_MS=200;
 const _STREAM_FADE_MAX_MS=350;
 const _STREAM_FADE_STAGGER_MS=16;
 const _STREAM_FADE_DONE_MAX_MS=320;
+const _STREAM_FADE_DONE_DRAIN_MAX_MS=900;
 const performance={performance_stub};
 {helpers}
 """
@@ -178,6 +179,20 @@ def test_stream_fade_uses_incremental_renderer_without_changing_default_path():
     assert "_wrapStreamingFadeWords" not in MESSAGES_JS
 
 
+def test_stream_fade_done_drain_has_hard_cap_for_large_buffered_responses():
+    drain_block = function_block(MESSAGES_JS, "_drainStreamFadeBeforeDone")
+    assert "const _STREAM_FADE_DONE_DRAIN_MAX_MS=900" in MESSAGES_JS
+    assert_contains_all(
+        drain_block,
+        [
+            "const drainStartedAt=performance.now();",
+            "performance.now()-drainStartedAt>=_STREAM_FADE_DONE_DRAIN_MAX_MS",
+            "if(_smdParser) _smdEndParser();",
+            "onDone();",
+        ],
+    )
+
+
 def test_stream_fade_css_is_opacity_only_and_hides_live_cursor():
     fade_css = STYLE_CSS[STYLE_CSS.index("OpenWebUI-style streaming word fade") :]
     assert "filter:" not in STYLE_CSS[STYLE_CSS.index("OpenWebUI-style streaming word fade") :].split(