diff --git a/langchain-core/src/language_models/base.ts b/langchain-core/src/language_models/base.ts
index b06b9f21a8e8..904c5910f5c7 100644
--- a/langchain-core/src/language_models/base.ts
+++ b/langchain-core/src/language_models/base.ts
@@ -643,6 +643,14 @@ export abstract class BaseLanguageModel<
  */
 export interface TokenUsage {
   completionTokens?: number;
+  completionTokensDetails?: {
+    reasoningTokens?: number;
+    acceptedPredictionTokens?: number;
+    rejectedPredictionTokens?: number;
+  };
   promptTokens?: number;
+  promptTokensDetails?: {
+    cachedTokens?: number;
+  };
   totalTokens?: number;
 }
diff --git a/libs/langchain-azure-openai/src/chat_models.ts b/libs/langchain-azure-openai/src/chat_models.ts
index 9b899d3c7889..2415756b3bda 100644
--- a/libs/langchain-azure-openai/src/chat_models.ts
+++ b/libs/langchain-azure-openai/src/chat_models.ts
@@ -696,7 +696,15 @@ export class AzureChatOpenAI
       {
         tokenUsage: {
           completionTokens: 0,
+          completionTokensDetails: {
+            reasoningTokens: 0,
+            acceptedPredictionTokens: 0,
+            rejectedPredictionTokens: 0,
+          },
           promptTokens: 0,
+          promptTokensDetails: {
+            cachedTokens: 0,
+          },
           totalTokens: 0,
         },
       }
diff --git a/libs/langchain-openai/src/llms.ts b/libs/langchain-openai/src/llms.ts
index 9bc4dd1cf1b1..b12fa9b1c18f 100644
--- a/libs/langchain-openai/src/llms.ts
+++ b/libs/langchain-openai/src/llms.ts
@@ -24,7 +24,15 @@ export type { OpenAICallOptions, OpenAIInput };
  */
 interface TokenUsage {
   completionTokens?: number;
+  completionTokensDetails?: {
+    reasoningTokens?: number;
+    acceptedPredictionTokens?: number;
+    rejectedPredictionTokens?: number;
+  };
   promptTokens?: number;
+  promptTokensDetails?: {
+    cachedTokens?: number;
+  };
   totalTokens?: number;
 }
 
@@ -342,13 +350,17 @@ export class OpenAI<CallOptions extends OpenAICallOptions = OpenAICallOptions>
       choices.push(...data.choices);
       const {
         completion_tokens: completionTokens,
+        completion_tokens_details: completionTokensDetails,
         prompt_tokens: promptTokens,
+        prompt_tokens_details: promptTokensDetails,
         total_tokens: totalTokens,
       } = data.usage
         ? data.usage
         : {
             completion_tokens: undefined,
+            completion_tokens_details: undefined,
             prompt_tokens: undefined,
+            prompt_tokens_details: undefined,
             total_tokens: undefined,
           };
 
@@ -357,10 +369,38 @@ export class OpenAI<CallOptions extends OpenAICallOptions = OpenAICallOptions>
           (tokenUsage.completionTokens ?? 0) + completionTokens;
       }
 
+      if (completionTokensDetails) {
+        if (!tokenUsage.completionTokensDetails) {
+          tokenUsage.completionTokensDetails = {};
+        }
+        if (completionTokensDetails.reasoning_tokens) {
+          tokenUsage.completionTokensDetails.reasoningTokens =
+            (tokenUsage.completionTokensDetails.reasoningTokens ?? 0) + completionTokensDetails.reasoning_tokens;
+        }
+        if (completionTokensDetails.accepted_prediction_tokens) {
+          tokenUsage.completionTokensDetails.acceptedPredictionTokens =
+            (tokenUsage.completionTokensDetails.acceptedPredictionTokens ?? 0) + completionTokensDetails.accepted_prediction_tokens;
+        }
+        if (completionTokensDetails.rejected_prediction_tokens) {
+          tokenUsage.completionTokensDetails.rejectedPredictionTokens =
+            (tokenUsage.completionTokensDetails.rejectedPredictionTokens ?? 0) + completionTokensDetails.rejected_prediction_tokens;
+        }
+      }
+
       if (promptTokens) {
         tokenUsage.promptTokens = (tokenUsage.promptTokens ?? 0) + promptTokens;
       }
 
+      if (promptTokensDetails) {
+        if (!tokenUsage.promptTokensDetails) {
+          tokenUsage.promptTokensDetails = {};
+        }
+        if (promptTokensDetails.cached_tokens) {
+          tokenUsage.promptTokensDetails.cachedTokens =
+            (tokenUsage.promptTokensDetails.cachedTokens ?? 0) + promptTokensDetails.cached_tokens;
+        }
+      }
+
       if (totalTokens) {
         tokenUsage.totalTokens = (tokenUsage.totalTokens ?? 0) + totalTokens;
       }