diff --git a/src/llm/cache.rs b/src/llm/cache.rs
index 087d2e400..cb753c8ab 100644
--- a/src/llm/cache.rs
+++ b/src/llm/cache.rs
@@ -632,10 +632,39 @@ impl EmbeddingService for LocalEmbeddingService {
             .send()
             .await?;
 
-        let result: Value = response.json().await?;
+        let status = response.status();
+        let response_text = response.text().await?;
+
+        if !status.is_success() {
+            debug!(
+                "Embedding service HTTP error {}: {}",
+                status,
+                response_text
+            );
+            return Err(format!(
+                "Embedding service returned HTTP {}: {}",
+                status,
+                response_text
+            ).into());
+        }
+
+        let result: Value = serde_json::from_str(&response_text)
+            .map_err(|e| {
+                debug!("Failed to parse embedding JSON: {} - Response: {}", e, response_text);
+                format!("Failed to parse embedding response JSON: {} - Response: {}", e, response_text)
+            })?;
+
+        if let Some(error) = result.get("error") {
+            debug!("Embedding service returned error: {}", error);
+            return Err(format!("Embedding service error: {}", error).into());
+        }
+
         let embedding = result["data"][0]["embedding"]
             .as_array()
-            .ok_or("Invalid embedding response")?
+            .ok_or_else(|| {
+                debug!("Invalid embedding response format. Expected data[0].embedding array. Got: {}", response_text);
+                format!("Invalid embedding response format - Expected data[0].embedding array, got: {}", response_text)
+            })?
             .iter()
             .filter_map(|v| v.as_f64().map(|f| f as f32))
             .collect();
diff --git a/src/main_module/bootstrap.rs b/src/main_module/bootstrap.rs
index cf4a0ca12..85a984d4c 100644
--- a/src/main_module/bootstrap.rs
+++ b/src/main_module/bootstrap.rs
@@ -748,17 +748,27 @@ fn init_llm_provider(
         let embedding_model = config_manager
             .get_config(&bot_id, "embedding-model", Some("all-MiniLM-L6-v2"))
             .unwrap_or_else(|_| "all-MiniLM-L6-v2".to_string());
+        let semantic_cache_enabled = config_manager
+            .get_config(&bot_id, "semantic-cache-enabled", Some("true"))
+            .unwrap_or_else(|_| "true".to_string())
+            .to_lowercase() == "true";
+
         info!("Embedding URL: {}", embedding_url);
         info!("Embedding Model: {}", embedding_model);
+        info!("Semantic Cache Enabled: {}", semantic_cache_enabled);
 
-        let embedding_service = Some(Arc::new(LocalEmbeddingService::new(
-            embedding_url,
-            embedding_model,
-        )) as Arc<dyn EmbeddingService>);
+        let embedding_service = if semantic_cache_enabled {
+            Some(Arc::new(LocalEmbeddingService::new(
+                embedding_url,
+                embedding_model,
+            )) as Arc<dyn EmbeddingService>)
+        } else {
+            None
+        };
 
         let cache_config = CacheConfig {
             ttl: 3600,
-            semantic_matching: true,
+            semantic_matching: semantic_cache_enabled,
             similarity_threshold: 0.85,
             max_similarity_checks: 100,
             key_prefix: "llm_cache".to_string(),