diff --git a/crates/tabby/src/serve.rs b/crates/tabby/src/serve.rs
index 3b1036603560..43ee77aa715e 100644
--- a/crates/tabby/src/serve.rs
+++ b/crates/tabby/src/serve.rs
@@ -5,7 +5,7 @@ use clap::Args;
 use hyper::StatusCode;
 use tabby_common::{
     api::{self, code::CodeSearch, event::EventLogger},
-    config::{Config, ConfigAccess, ModelConfig, StaticConfigAccess},
+    config::{Config, ConfigAccess, LocalModelConfig, ModelConfig, StaticConfigAccess},
     usage,
 };
 use tabby_inference::Embedding;
@@ -398,5 +398,13 @@ fn merge_args(config: &Config, args: &ServeArgs) -> Config {
         ));
     }
 
+    if config.model.embedding.is_none() {
+        config.model.embedding = Some(ModelConfig::Local(LocalModelConfig {
+            model_id: "Nomic-Embed-Text".to_string(),
+            parallelism: 4,
+            num_gpu_layers: 9999,
+        }))
+    }
+
     config
 }
diff --git a/crates/tabby/src/services/embedding.rs b/crates/tabby/src/services/embedding.rs
index d93f4e38dba3..e515adee0d12 100644
--- a/crates/tabby/src/services/embedding.rs
+++ b/crates/tabby/src/services/embedding.rs
@@ -6,17 +6,5 @@ use tabby_inference::Embedding;
 use super::model;
 
 pub async fn create(config: Option<&ModelConfig>) -> Arc<dyn Embedding> {
-    if let Some(config) = config {
-        model::load_embedding(config).await
-    } else {
-        model::load_embedding(&default_config()).await
-    }
-}
-
-fn default_config() -> ModelConfig {
-    ModelConfig::Local(LocalModelConfig {
-        model_id: "Nomic-Embed-Text".to_string(),
-        parallelism: 4,
-        num_gpu_layers: 9999,
-    })
+    model::load_embedding(config.expect("Embedding model is not specified")).await
 }