diff --git a/ansible/roles/turboquant-deploy/tasks/integration_test.yml b/ansible/roles/turboquant-deploy/tasks/integration_test.yml
new file mode 100644
index 00000000..a39e1ef0
--- /dev/null
+++ b/ansible/roles/turboquant-deploy/tasks/integration_test.yml
@@ -0,0 +1,38 @@
+---
+# Post-deploy integration tests
+
+- name: Wait for TurboQuant service to be ready
+  uri:
+    url: "http://localhost:{{ llama_cpp_port }}/v1/models"
+    method: GET
+    status_code: 200
+  register: model_check
+  retries: 30
+  delay: 5
+  until: model_check.status == 200
+
+- name: Test inference with tool call
+  uri:
+    url: "http://localhost:{{ llama_cpp_port }}/v1/chat/completions"
+    method: POST
+    body_format: json
+    body:
+      model: "gemma-4"
+      messages:
+        - role: "user"
+          content: "Say 'test passed' and nothing else."
+      max_tokens: 20
+    status_code: 200
+  register: inference_test
+
+- name: Verify inference response
+  assert:
+    that:
+      - "'choices' in inference_test.json"
+      - "inference_test.json.choices | length > 0"
+    fail_msg: "Inference test failed - no valid response"
+    success_msg: "Inference test passed"
+
+- name: Report test results
+  debug:
+    msg: "Integration test passed on {{ inventory_hostname }}"