diff --git a/contrib/spring-ai/src/main/java/com/google/adk/models/springai/MessageConverter.java b/contrib/spring-ai/src/main/java/com/google/adk/models/springai/MessageConverter.java
index 3983b08a5..ab10a6521 100644
--- a/contrib/spring-ai/src/main/java/com/google/adk/models/springai/MessageConverter.java
+++ b/contrib/spring-ai/src/main/java/com/google/adk/models/springai/MessageConverter.java
@@ -22,6 +22,7 @@
 import com.google.adk.models.LlmResponse;
 import com.google.genai.types.Content;
 import com.google.genai.types.FunctionCall;
+import com.google.genai.types.GenerateContentResponseUsageMetadata;
 import com.google.genai.types.Part;
 import java.net.URI;
 import java.util.ArrayList;
@@ -32,6 +33,8 @@
 import org.springframework.ai.chat.messages.SystemMessage;
 import org.springframework.ai.chat.messages.ToolResponseMessage;
 import org.springframework.ai.chat.messages.UserMessage;
+import org.springframework.ai.chat.metadata.EmptyUsage;
+import org.springframework.ai.chat.metadata.Usage;
 import org.springframework.ai.chat.model.ChatResponse;
 import org.springframework.ai.chat.model.Generation;
 import org.springframework.ai.chat.prompt.ChatOptions;
@@ -318,11 +321,27 @@ public LlmResponse toLlmResponse(ChatResponse chatResponse, boolean isStreaming)
     boolean isPartial = isStreaming && isPartialResponse(assistantMessage);
     boolean isTurnComplete = !isStreaming || isTurnCompleteResponse(chatResponse);
 
-    return LlmResponse.builder()
-        .content(content)
-        .partial(isPartial)
-        .turnComplete(isTurnComplete)
-        .build();
+    LlmResponse.Builder responseBuilder =
+        LlmResponse.builder().content(content).partial(isPartial).turnComplete(isTurnComplete);
+
+    if (chatResponse.getMetadata() != null
+        && chatResponse.getMetadata().getUsage() != null
+        && !(chatResponse.getMetadata().getUsage() instanceof EmptyUsage)) {
+      Usage springUsage = chatResponse.getMetadata().getUsage();
+
+      GenerateContentResponseUsageMetadata adkUsage =
+          GenerateContentResponseUsageMetadata.builder()
+              .promptTokenCount(nullSafeInt(springUsage.getPromptTokens()))
+              .candidatesTokenCount(nullSafeInt(springUsage.getCompletionTokens()))
+              .totalTokenCount(nullSafeInt(springUsage.getTotalTokens()))
+              .build();
+      responseBuilder.usageMetadata(adkUsage);
+    }
+    return responseBuilder.build();
+  }
+
+  private int nullSafeInt(Integer value) {
+    return value != null ? value.intValue() : 0;
   }
 
   /** Determines if an assistant message represents a partial response in streaming. */
diff --git a/contrib/spring-ai/src/test/java/com/google/adk/models/springai/MessageConverterTest.java b/contrib/spring-ai/src/test/java/com/google/adk/models/springai/MessageConverterTest.java
index b861a71f2..513b61179 100644
--- a/contrib/spring-ai/src/test/java/com/google/adk/models/springai/MessageConverterTest.java
+++ b/contrib/spring-ai/src/test/java/com/google/adk/models/springai/MessageConverterTest.java
@@ -16,7 +16,8 @@
 package com.google.adk.models.springai;
 
 import static org.assertj.core.api.Assertions.assertThat;
-import static org.junit.jupiter.api.Assertions.assertThrows;
+import static org.junit.jupiter.api.Assertions.*;
+import static org.junit.jupiter.api.Assertions.assertEquals;
 
 import com.fasterxml.jackson.databind.ObjectMapper;
 import com.google.adk.models.LlmRequest;
@@ -33,6 +34,8 @@
 import org.springframework.ai.chat.messages.Message;
 import org.springframework.ai.chat.messages.SystemMessage;
 import org.springframework.ai.chat.messages.UserMessage;
+import org.springframework.ai.chat.metadata.ChatResponseMetadata;
+import org.springframework.ai.chat.metadata.DefaultUsage;
 import org.springframework.ai.chat.model.ChatResponse;
 import org.springframework.ai.chat.model.Generation;
 import org.springframework.ai.chat.prompt.Prompt;
@@ -237,6 +240,70 @@ void testToLlmResponseFromChatResponseWithToolCalls() {
     assertThat(functionCallPart.functionCall().get().id()).contains("call_123");
   }
 
+  @Test
+  void testUsageMetadataShouldBeEmptyWhenSpringAiMetadataIsNull() {
+    MessageConverter converter = new MessageConverter(new ObjectMapper());
+    AssistantMessage assistantMessage = new AssistantMessage("intermediate chunk");
+    Generation generation = new Generation(assistantMessage);
+
+    ChatResponse chatResponse = new ChatResponse(List.of(generation), null);
+
+    LlmResponse llmResponse = converter.toLlmResponse(chatResponse, true);
+
+    assertTrue(
+        llmResponse.usageMetadata().isEmpty(),
+        "Expected usageMetadata to be empty for intermediate stream chunks lacking metadata");
+  }
+
+  @Test
+  void testUsageMetadataShouldBeEmptyWhenSpringAiUsageIsNull() {
+    MessageConverter converter = new MessageConverter(new ObjectMapper());
+    AssistantMessage assistantMessage = new AssistantMessage("intermediate chunk");
+    Generation generation = new Generation(assistantMessage);
+
+    ChatResponseMetadata metadata = ChatResponseMetadata.builder().id("resp-no-usage").build();
+
+    ChatResponse chatResponse = new ChatResponse(List.of(generation), metadata);
+
+    LlmResponse llmResponse = converter.toLlmResponse(chatResponse, true);
+
+    assertTrue(
+        llmResponse.usageMetadata().isEmpty(),
+        "Expected usageMetadata to be empty when metadata exists but usage is null");
+  }
+
+  @Test
+  void testUsageMetadataShouldDefaultToZeroWhenSpringAiTokensAreNull() {
+    MessageConverter converter = new MessageConverter(new ObjectMapper());
+    AssistantMessage assistantMessage = new AssistantMessage("final chunk");
+    Generation generation = new Generation(assistantMessage);
+
+    // Anonymous implementation to simulate incomplete provider data where some token counts are
+    // null
+    DefaultUsage incompleteUsage = new DefaultUsage(null, null, 42);
+    ChatResponseMetadata metadata =
+        ChatResponseMetadata.builder().id("resp-partial-tokens").usage(incompleteUsage).build();
+
+    ChatResponse chatResponse = new ChatResponse(List.of(generation), metadata);
+
+    LlmResponse llmResponse = converter.toLlmResponse(chatResponse, false);
+
+    assertTrue(llmResponse.usageMetadata().isPresent(), "Expected usageMetadata to be present");
+
+    assertEquals(
+        0,
+        llmResponse.usageMetadata().get().promptTokenCount().orElse(-1),
+        "Null prompt tokens should default to 0");
+    assertEquals(
+        0,
+        llmResponse.usageMetadata().get().candidatesTokenCount().orElse(-1),
+        "Null completion tokens should default to 0");
+    assertEquals(
+        42,
+        llmResponse.usageMetadata().get().totalTokenCount().orElse(-1),
+        "Total tokens should be mapped correctly");
+  }
+
   @Test
   void testToolCallIdPreservedInConversion() {
     // Create AssistantMessage with tool call including ID