diff --git a/crates/meilisearch-types/src/error.rs b/crates/meilisearch-types/src/error.rs
index 859563d8a..6c547d51e 100644
--- a/crates/meilisearch-types/src/error.rs
+++ b/crates/meilisearch-types/src/error.rs
@@ -454,7 +454,10 @@ impl ErrorCode for milli::Error {
                     }
                     UserError::CriterionError(_) => Code::InvalidSettingsRankingRules,
                     UserError::InvalidGeoField { .. } => Code::InvalidDocumentGeoField,
-                    UserError::InvalidVectorDimensions { .. } => Code::InvalidVectorDimensions,
+                    UserError::InvalidVectorDimensions { .. }
+                    | UserError::InvalidIndexingVectorDimensions { .. } => {
+                        Code::InvalidVectorDimensions
+                    }
                     UserError::InvalidVectorsMapType { .. }
                     | UserError::InvalidVectorsEmbedderConf { .. } => Code::InvalidVectorsType,
                     UserError::TooManyVectors(_, _) => Code::TooManyVectors,
diff --git a/crates/meilisearch/tests/vector/mod.rs b/crates/meilisearch/tests/vector/mod.rs
index 67da51702..5e34a4c23 100644
--- a/crates/meilisearch/tests/vector/mod.rs
+++ b/crates/meilisearch/tests/vector/mod.rs
@@ -164,6 +164,87 @@ async fn add_remove_user_provided() {
     "###);
 }
 
+#[actix_rt::test]
+async fn user_provide_mismatched_embedding_dimension() {
+    let server = Server::new().await;
+    let index = server.index("doggo");
+
+    let (response, code) = index
+        .update_settings(json!({
+          "embedders": {
+              "manual": {
+                  "source": "userProvided",
+                  "dimensions": 3,
+              }
+          },
+        }))
+        .await;
+    snapshot!(code, @"202 Accepted");
+    server.wait_task(response.uid()).await.succeeded();
+
+    let documents = json!([
+      {"id": 0, "name": "kefir", "_vectors": { "manual": [0, 0] }},
+    ]);
+    let (value, code) = index.add_documents(documents, None).await;
+    snapshot!(code, @"202 Accepted");
+    let task = index.wait_task(value.uid()).await;
+    snapshot!(task, @r###"
+    {
+      "uid": "[uid]",
+      "batchUid": "[batch_uid]",
+      "indexUid": "doggo",
+      "status": "failed",
+      "type": "documentAdditionOrUpdate",
+      "canceledBy": null,
+      "details": {
+        "receivedDocuments": 1,
+        "indexedDocuments": 0
+      },
+      "error": {
+        "message": "Index `doggo`: Invalid vector dimensions in document with id `0` in `._vectors.manual`.\n  - note: embedding #0 has dimensions 2\n  - note: embedder `manual` requires 3",
+        "code": "invalid_vector_dimensions",
+        "type": "invalid_request",
+        "link": "https://docs.meilisearch.com/errors#invalid_vector_dimensions"
+      },
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
+
+    let new_document = json!([
+      {"id": 0, "name": "kefir", "_vectors": { "manual": [[0, 0], [1, 1], [2, 2]] }},
+    ]);
+    let (response, code) = index.add_documents(new_document, None).await;
+    snapshot!(code, @"202 Accepted");
+    let task = index.wait_task(response.uid()).await;
+    snapshot!(task, @r###"
+    {
+      "uid": "[uid]",
+      "batchUid": "[batch_uid]",
+      "indexUid": "doggo",
+      "status": "failed",
+      "type": "documentAdditionOrUpdate",
+      "canceledBy": null,
+      "details": {
+        "receivedDocuments": 1,
+        "indexedDocuments": 0
+      },
+      "error": {
+        "message": "Index `doggo`: Invalid vector dimensions in document with id `0` in `._vectors.manual`.\n  - note: embedding #0 has dimensions 2\n  - note: embedder `manual` requires 3",
+        "code": "invalid_vector_dimensions",
+        "type": "invalid_request",
+        "link": "https://docs.meilisearch.com/errors#invalid_vector_dimensions"
+      },
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
+}
+
 async fn generate_default_user_provided_documents(server: &Server) -> Index {
     let index = server.index("doggo");
 
diff --git a/crates/milli/src/error.rs b/crates/milli/src/error.rs
index e1098cfa5..e61283e4c 100644
--- a/crates/milli/src/error.rs
+++ b/crates/milli/src/error.rs
@@ -129,6 +129,14 @@ and can not be more than 511 bytes.", .document_id.to_string()
     InvalidGeoField(#[from] GeoError),
     #[error("Invalid vector dimensions: expected: `{}`, found: `{}`.", .expected, .found)]
     InvalidVectorDimensions { expected: usize, found: usize },
+    #[error("Invalid vector dimensions in document with id `{document_id}` in `._vectors.{embedder_name}`.\n  - note: embedding #{embedding_index} has dimensions {found}\n  - note: embedder `{embedder_name}` requires {expected}")]
+    InvalidIndexingVectorDimensions {
+        embedder_name: String,
+        document_id: String,
+        embedding_index: usize,
+        expected: usize,
+        found: usize,
+    },
     #[error("The `_vectors` field in the document with id: `{document_id}` is not an object. Was expecting an object with a key for each embedder with manually provided vectors, but instead got `{value}`")]
     InvalidVectorsMapType { document_id: String, value: Value },
     #[error("Bad embedder configuration in the document with id: `{document_id}`. {error}")]
diff --git a/crates/milli/src/update/new/extract/vectors/mod.rs b/crates/milli/src/update/new/extract/vectors/mod.rs
index 6820ee67b..696864e7f 100644
--- a/crates/milli/src/update/new/extract/vectors/mod.rs
+++ b/crates/milli/src/update/new/extract/vectors/mod.rs
@@ -121,6 +121,7 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
                             // do we have set embeddings?
                             if let Some(embeddings) = new_vectors.embeddings {
                                 chunks.set_vectors(
+                                    update.external_document_id(),
                                     update.docid(),
                                     embeddings
                                         .into_vec(&context.doc_alloc, embedder_name)
@@ -128,7 +129,7 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
                                             document_id: update.external_document_id().to_string(),
                                             error: error.to_string(),
                                         })?,
-                                );
+                                )?;
                             } else if new_vectors.regenerate {
                                 let new_rendered = prompt.render_document(
                                     update.external_document_id(),
@@ -209,6 +210,7 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
                             chunks.set_regenerate(insertion.docid(), new_vectors.regenerate);
                             if let Some(embeddings) = new_vectors.embeddings {
                                 chunks.set_vectors(
+                                    insertion.external_document_id(),
                                     insertion.docid(),
                                     embeddings
                                         .into_vec(&context.doc_alloc, embedder_name)
@@ -218,7 +220,7 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
                                                 .to_string(),
                                             error: error.to_string(),
                                         })?,
-                                );
+                                )?;
                             } else if new_vectors.regenerate {
                                 let rendered = prompt.render_document(
                                     insertion.external_document_id(),
@@ -273,6 +275,7 @@ struct Chunks<'a, 'b, 'extractor> {
     embedder: &'a Embedder,
     embedder_id: u8,
     embedder_name: &'a str,
+    dimensions: usize,
     prompt: &'a Prompt,
     possible_embedding_mistakes: &'a PossibleEmbeddingMistakes,
     user_provided: &'a RefCell<EmbeddingExtractorData<'extractor>>,
@@ -297,6 +300,7 @@ impl<'a, 'b, 'extractor> Chunks<'a, 'b, 'extractor> {
         let capacity = embedder.prompt_count_in_chunk_hint() * embedder.chunk_count_hint();
         let texts = BVec::with_capacity_in(capacity, doc_alloc);
         let ids = BVec::with_capacity_in(capacity, doc_alloc);
+        let dimensions = embedder.dimensions();
         Self {
             texts,
             ids,
@@ -309,6 +313,7 @@ impl<'a, 'b, 'extractor> Chunks<'a, 'b, 'extractor> {
             embedder_name,
             user_provided,
             has_manual_generation: None,
+            dimensions,
         }
     }
 
@@ -490,7 +495,25 @@ impl<'a, 'b, 'extractor> Chunks<'a, 'b, 'extractor> {
         }
     }
 
-    fn set_vectors(&self, docid: DocumentId, embeddings: Vec<Embedding>) {
+    fn set_vectors(
+        &self,
+        external_docid: &'a str,
+        docid: DocumentId,
+        embeddings: Vec<Embedding>,
+    ) -> Result<()> {
+        for (embedding_index, embedding) in embeddings.iter().enumerate() {
+            if embedding.len() != self.dimensions {
+                return Err(UserError::InvalidIndexingVectorDimensions {
+                    expected: self.dimensions,
+                    found: embedding.len(),
+                    embedder_name: self.embedder_name.to_string(),
+                    document_id: external_docid.to_string(),
+                    embedding_index,
+                }
+                .into());
+            }
+        }
         self.sender.set_vectors(docid, self.embedder_id, embeddings).unwrap();
+        Ok(())
     }
 }