diff --git a/crates/meilisearch/tests/search/hybrid.rs b/crates/meilisearch/tests/search/hybrid.rs
index 3282a357a0..ea13c3048b 100644
--- a/crates/meilisearch/tests/search/hybrid.rs
+++ b/crates/meilisearch/tests/search/hybrid.rs
@@ -625,3 +625,90 @@ async fn retrieve_vectors() {
     ]
     "###);
 }
+
+#[actix_rt::test]
+async fn reproduce_hybrid_distinct_bug() {
+    let server = Server::new().await;
+    let index = server.index("test");
+
+    // Set up embedder with higher dimensions for more extreme differences
+    let (response, code) = index
+        .update_settings(json!({
+            "embedders": {
+                "default": {
+                    "source": "userProvided",
+                    "dimensions": 8  // Increased from 2 to 8 dimensions
+                }
+            }
+        }))
+        .await;
+    assert_eq!(202, code);
+    index.wait_task(response.uid()).await.succeeded();
+
+    // Documents with IDENTICAL distinct values but extreme content differences
+    let test_documents = json!([
+        {
+            "id": 1,
+            "title": "red nike running shoes athletic footwear",  // Strong keyword match
+            "product_id": "DUPLICATE_PRODUCT_ID",
+            "_vectors": {"default": [-1.0, -1.0, -1.0, -1.0, -1.0, -1.0, -1.0, -1.0]}  // Extreme low similarity
+        },
+        {
+            "id": 2,
+            "title": "blue adidas casual shirt clothing apparel",  // Completely different keywords
+            "product_id": "DUPLICATE_PRODUCT_ID",  // SAME product_id!
+            "_vectors": {"default": [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]}   // Extreme high similarity
+        }
+    ]);
+
+    let (response, code) = index.add_documents(test_documents, Some("id")).await;
+    assert_eq!(202, code);
+    index.wait_task(response.uid()).await.succeeded();
+
+    // Set distinct on product_id
+    let (task, _) = index.update_distinct_attribute(json!("product_id")).await;
+    index.wait_task(task.uid()).await.succeeded();
+
+    // Hybrid search - should trigger both vector and keyword search
+    let (response, code) = index
+        .search_post(json!({
+            "q": "red nike running shoes",      // Should only match doc 1 well in keyword search
+            "vector": [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0],       // Should only match doc 2 well in vector search
+            "hybrid": {
+                "embedder": "default",
+                "semanticRatio": 0.5
+            },
+            "limit": 10
+        }))
+        .await;
+
+    assert_eq!(200, code);
+    let hits = response["hits"].as_array().unwrap();
+
+    println!("SEARCH RESULTS:");
+    for (i, hit) in hits.iter().enumerate() {
+        println!("  Hit {}: id={}, product_id={}", i, hit["id"], hit["product_id"]);
+    }
+
+    // Check for duplicate product_ids - THIS IS THE BUG
+    let product_ids: Vec<&str> = hits.iter().filter_map(|hit| hit["product_id"].as_str()).collect();
+
+    let unique_count = product_ids.iter().collect::<std::collections::HashSet<_>>().len();
+
+    if product_ids.len() > unique_count {
+        panic!(
+            "BUG REPRODUCED! Found {} total results but only {} unique product_ids. \
+               This means distinct filtering failed in hybrid search! Results: {:?}",
+            product_ids.len(),
+            unique_count,
+            product_ids
+        );
+    }
+
+    // If only one result, distinct filtering worked (but we hoped to reproduce the bug)
+    if hits.len() == 1 {
+        println!("Only one result returned - distinct filtering working or different issue");
+    } else {
+        println!("Multiple results with unique product_ids - distinct filtering appears to work");
+    }
+}