diff --git a/litellm-rust/crates/gateway-inference/src/chat_completions.rs b/litellm-rust/crates/gateway-inference/src/chat_completions.rs
index 41a3ac0fbc9..87380ead495 100644
--- a/litellm-rust/crates/gateway-inference/src/chat_completions.rs
+++ b/litellm-rust/crates/gateway-inference/src/chat_completions.rs
@@ -187,23 +187,30 @@ fn split(mut body: Map<String, Value>) -> (Value, Map<String, Value>) {
 
 /// A deployment's connection for a provider chat translates. Python's router reads each
 /// `os.environ/` setting from the environment when it loads the deployment, and `completion`
-/// hands the Vertex AI settings to the call, where a request body cannot name them.
+/// hands the Vertex AI and AWS settings to the call, where a request body cannot name them.
 async fn connection(
     gateway: &Gateway,
     deployment: &Deployment,
     optional_params: &mut Map<String, Value>,
 ) -> Result<(Option<String>, Option<String>), Error> {
-    let vertex = match serde_json::to_value(&deployment.litellm_params.vertex) {
-        Ok(Value::Object(fields)) => fields,
-        _ => Map::new(),
-    };
+    // The Vertex AI and AWS settings both live in their own groups of `litellm_params`, not in
+    // `extra`, so each is copied into the call's params, where the provider config reads it.
+    let mut settings = Map::new();
+    for group in [
+        serde_json::to_value(&deployment.litellm_params.vertex),
+        serde_json::to_value(&deployment.litellm_params.aws),
+    ] {
+        if let Ok(Value::Object(fields)) = group {
+            settings.extend(fields);
+        }
+    }
     let references: Vec<&str> = [
         deployment.api_key.as_deref(),
         deployment.api_base.as_deref(),
     ]
     .into_iter()
     .flatten()
-    .chain(vertex.values().filter_map(Value::as_str))
+    .chain(settings.values().filter_map(Value::as_str))
     .filter_map(|value| value.strip_prefix(ENV_REFERENCE))
     .collect();
     let secrets = match references.is_empty() {
@@ -221,7 +228,7 @@ async fn connection(
         Some(name) => secrets.as_ref().and_then(|secrets| secrets.get(name)),
         None => Some(value.to_owned()),
     };
-    for (name, value) in vertex {
+    for (name, value) in settings {
         let value = match value {
             Value::String(text) => resolve(&text).map(Value::String),
             Value::Null => None,
diff --git a/litellm-rust/crates/gateway/tests/chat_translation.rs b/litellm-rust/crates/gateway/tests/chat_translation.rs
index aa0c1da8f6f..3c7f7acf305 100644
--- a/litellm-rust/crates/gateway/tests/chat_translation.rs
+++ b/litellm-rust/crates/gateway/tests/chat_translation.rs
@@ -276,6 +276,13 @@ async fn harness() -> Harness {
       model: bedrock/{BEDROCK_MODEL}
       api_base: {upstream}
       api_key: bedrock-token
+  - model_name: bedrock-keys
+    litellm_params:
+      model: bedrock/{BEDROCK_MODEL}
+      aws_access_key_id: AKIACONFIGKEY
+      aws_secret_access_key: config-secret
+      aws_region_name: us-west-2
+      aws_bedrock_runtime_endpoint: {upstream}
   - model_name: claude-dropping
     litellm_params:
       model: anthropic/{ANTHROPIC_MODEL}
@@ -449,6 +456,24 @@ async fn anthropic_tool_round_trip() {
     );
 }
 
+#[tokio::test]
+async fn bedrock_signs_with_the_keys_region_and_endpoint_in_the_config() {
+    let harness = harness().await;
+    harness
+        .json(json!({"model": "bedrock-keys", "messages": answered("tooluse_1"), "max_tokens": 16}))
+        .await;
+
+    let sent = harness.sent();
+    assert_eq!(sent.len(), 1, "the configured endpoint gets the call");
+    assert!(sent[0].path.ends_with("/converse"), "{}", sent[0].path);
+    let authorization = sent[0].header("authorization").unwrap();
+    assert!(
+        authorization.starts_with("AWS4-HMAC-SHA256 Credential=AKIACONFIGKEY/")
+            && authorization.contains("/us-west-2/bedrock/aws4_request"),
+        "{authorization}"
+    );
+}
+
 #[tokio::test]
 async fn bedrock_tool_round_trip() {
     let harness = harness().await;