diff --git a/src/lib/i18n/locales/de-DE/translation.json b/src/lib/i18n/locales/de-DE/translation.json
index 3dd27eca04..6bfd8624a8 100644
--- a/src/lib/i18n/locales/de-DE/translation.json
+++ b/src/lib/i18n/locales/de-DE/translation.json
@@ -2616,9 +2616,9 @@
 	"settings.admin.documents.title": "Dokumente",
 	"settings.admin.documents.tokenizerModel.description": "Tokenizer-Modell für die tokenbasierte Aufteilung mit Transformern.",
 	"settings.admin.documents.tokenizerModel.label": "Tokenizer-Modell",
-	"settings.admin.documents.topK.description": "Maximale Anzahl abgerufener Chunks, die an das Modell übergeben werden.",
+	"settings.admin.documents.topK.description": "Maximale Anzahl der Chunks, die an das Modell übergeben werden. Bei hybrider Suche auch die Anzahl der pro Anfrage abgerufenen Kandidaten für den Reranker.",
 	"settings.admin.documents.topK.label": "Top-K",
-	"settings.admin.documents.topKReranker.description": "Maximale Anzahl an Ergebnissen der hybriden Suche, die an den Reranker gesendet werden.",
+	"settings.admin.documents.topKReranker.description": "Maximale Anzahl der Chunks, die pro Anfrage nach dem Reranking behalten und an das Modell übergeben werden.",
 	"settings.admin.documents.topKReranker.label": "Top-K-Reranker",
 	"settings.admin.documents.updateEmbeddingModel.label": "Embedding-Modell aktualisieren",
 	"settings.admin.documents.useBase64.description": "Sendet PDFs als Base64-Daten-URLs, statt sie zuvor hochzuladen.",
diff --git a/src/lib/i18n/locales/dg-DG/translation.json b/src/lib/i18n/locales/dg-DG/translation.json
index 0731885f3d..c8f539f90b 100644
--- a/src/lib/i18n/locales/dg-DG/translation.json
+++ b/src/lib/i18n/locales/dg-DG/translation.json
@@ -2616,9 +2616,9 @@
 	"settings.admin.documents.title": "Many Documents",
 	"settings.admin.documents.tokenizerModel.description": "Tokenizer model used for transformer token splitting. Much brain.",
 	"settings.admin.documents.tokenizerModel.label": "Such Tokenizer Model",
-	"settings.admin.documents.topK.description": "Maximum number of retrieved chunks returned to the model. Much sniff.",
+	"settings.admin.documents.topK.description": "Maximum number of chunks passed to the model. With hybrid search, also the number of candidates retrieved per query for the reranker. Much sniff.",
 	"settings.admin.documents.topK.label": "Such Top K",
-	"settings.admin.documents.topKReranker.description": "Maximum number of hybrid results sent to the reranker. Very picky.",
+	"settings.admin.documents.topKReranker.description": "Maximum number of chunks kept per query after reranking and passed to the model. Very picky.",
 	"settings.admin.documents.topKReranker.label": "Such Top K Reranker",
 	"settings.admin.documents.updateEmbeddingModel.label": "Update such embedding model",
 	"settings.admin.documents.useBase64.description": "Send PDFs as base64 data URLs instead of uploading first. So swift.",
diff --git a/src/lib/i18n/locales/en-US/translation.json b/src/lib/i18n/locales/en-US/translation.json
index 7b81819a7b..97e3536a02 100644
--- a/src/lib/i18n/locales/en-US/translation.json
+++ b/src/lib/i18n/locales/en-US/translation.json
@@ -2616,9 +2616,9 @@
 	"settings.admin.documents.title": "Documents",
 	"settings.admin.documents.tokenizerModel.description": "Tokenizer model used for transformer token splitting.",
 	"settings.admin.documents.tokenizerModel.label": "Tokenizer Model",
-	"settings.admin.documents.topK.description": "Maximum number of retrieved chunks returned to the model.",
+	"settings.admin.documents.topK.description": "Maximum number of chunks passed to the model. With hybrid search, also the number of candidates retrieved per query for the reranker.",
 	"settings.admin.documents.topK.label": "Top K",
-	"settings.admin.documents.topKReranker.description": "Maximum number of hybrid results sent to the reranker.",
+	"settings.admin.documents.topKReranker.description": "Maximum number of chunks kept per query after reranking and passed to the model.",
 	"settings.admin.documents.topKReranker.label": "Top K Reranker",
 	"settings.admin.documents.updateEmbeddingModel.label": "Update embedding model",
 	"settings.admin.documents.useBase64.description": "Send PDFs as base64 data URLs instead of uploading first.",