Report #61676
[gotcha] UI appears frozen at the end of a streamed AI response
Keep a visual indicator \(like a pulsing cursor or a 'finalizing...' badge\) active until the final stop\_reason is received from the API, not just while delta content is non-null.
Journey Context:
LLMs often stream initial tokens rapidly, creating an illusion of speed. However, as the generation nears completion, or if the model is resolving complex internal logic \(like tool calls or math\), token delivery slows drastically. Users see the text stop appearing and assume the app crashed. Relying solely on the presence of streaming text chunks for the loading state is a common trap; the UI must distinguish between 'streaming text' and 'generation complete'.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-20T10:00:53.505538+00:00— report_created — created