diff --git a/packages/xai/src/xai-chat-language-model.test.ts b/packages/xai/src/xai-chat-language-model.test.ts
index 0a43feb10395..4c87215d3058 100644
--- a/packages/xai/src/xai-chat-language-model.test.ts
+++ b/packages/xai/src/xai-chat-language-model.test.ts
@@ -190,6 +190,7 @@ describe('XaiChatLanguageModel', () => {
 
       expect(usage).toMatchInlineSnapshot(`
         {
+          "cachedInputTokens": undefined,
           "inputTokens": 20,
           "outputTokens": 5,
           "reasoningTokens": undefined,
@@ -799,6 +800,7 @@ describe('XaiChatLanguageModel', () => {
             "finishReason": "stop",
             "type": "finish",
             "usage": {
+              "cachedInputTokens": undefined,
               "inputTokens": 4,
               "outputTokens": 32,
               "reasoningTokens": undefined,
@@ -862,6 +864,7 @@ describe('XaiChatLanguageModel', () => {
             "finishReason": "stop",
             "type": "finish",
             "usage": {
+              "cachedInputTokens": undefined,
               "inputTokens": 4,
               "outputTokens": 32,
               "reasoningTokens": undefined,
@@ -940,6 +943,7 @@ describe('XaiChatLanguageModel', () => {
             "finishReason": "tool-calls",
             "type": "finish",
             "usage": {
+              "cachedInputTokens": undefined,
               "inputTokens": 183,
               "outputTokens": 133,
               "reasoningTokens": undefined,
@@ -1126,6 +1130,7 @@ describe('XaiChatLanguageModel', () => {
             "finishReason": "stop",
             "type": "finish",
             "usage": {
+              "cachedInputTokens": undefined,
               "inputTokens": 4,
               "outputTokens": 30,
               "reasoningTokens": undefined,
@@ -1252,6 +1257,7 @@ describe('XaiChatLanguageModel', () => {
 
       expect(usage).toMatchInlineSnapshot(`
         {
+          "cachedInputTokens": undefined,
           "inputTokens": 15,
           "outputTokens": 20,
           "reasoningTokens": 10,
@@ -1334,6 +1340,7 @@ describe('XaiChatLanguageModel', () => {
             "finishReason": "stop",
             "type": "finish",
             "usage": {
+              "cachedInputTokens": undefined,
               "inputTokens": 15,
               "outputTokens": 20,
               "reasoningTokens": 10,
@@ -1424,6 +1431,7 @@ describe('XaiChatLanguageModel', () => {
             "finishReason": "stop",
             "type": "finish",
             "usage": {
+              "cachedInputTokens": undefined,
               "inputTokens": 15,
               "outputTokens": 20,
               "reasoningTokens": 10,
@@ -1567,6 +1575,7 @@ describe('doStream with raw chunks', () => {
           "finishReason": "stop",
           "type": "finish",
           "usage": {
+            "cachedInputTokens": undefined,
             "inputTokens": 10,
             "outputTokens": 5,
             "reasoningTokens": undefined,
diff --git a/packages/xai/src/xai-chat-language-model.ts b/packages/xai/src/xai-chat-language-model.ts
index 3a95d173dad4..29602f2ea085 100644
--- a/packages/xai/src/xai-chat-language-model.ts
+++ b/packages/xai/src/xai-chat-language-model.ts
@@ -294,6 +294,8 @@ export class XaiChatLanguageModel implements LanguageModelV3 {
         reasoningTokens:
           response.usage.completion_tokens_details?.reasoning_tokens ??
           undefined,
+        cachedInputTokens:
+          response.usage.prompt_tokens_details?.cached_tokens ?? undefined,
       },
       request: { body },
       response: {
@@ -333,6 +335,8 @@ export class XaiChatLanguageModel implements LanguageModelV3 {
       inputTokens: undefined,
       outputTokens: undefined,
       totalTokens: undefined,
+      reasoningTokens: undefined,
+      cachedInputTokens: undefined,
     };
     let isFirstChunk = true;
     const contentBlocks: Record<string, { type: 'text' | 'reasoning' }> = {};
@@ -392,6 +396,8 @@ export class XaiChatLanguageModel implements LanguageModelV3 {
               usage.reasoningTokens =
                 value.usage.completion_tokens_details?.reasoning_tokens ??
                 undefined;
+              usage.cachedInputTokens =
+                value.usage.prompt_tokens_details?.cached_tokens ?? undefined;
             }
 
             const choice = value.choices[0];
@@ -523,9 +529,20 @@ const xaiUsageSchema = z.object({
   prompt_tokens: z.number(),
   completion_tokens: z.number(),
   total_tokens: z.number(),
+  prompt_tokens_details: z
+    .object({
+      text_tokens: z.number().nullish(),
+      audio_tokens: z.number().nullish(),
+      image_tokens: z.number().nullish(),
+      cached_tokens: z.number().nullish(),
+    })
+    .nullish(),
   completion_tokens_details: z
     .object({
       reasoning_tokens: z.number().nullish(),
+      audio_tokens: z.number().nullish(),
+      accepted_prediction_tokens: z.number().nullish(),
+      rejected_prediction_tokens: z.number().nullish(),
     })
     .nullish(),
 });