fix typo and cached_input token

src/pages/Chatbox.tsx

@@ -237,17 +237,6 @@ export default function ChatBOX() {
 
   const _completeWithFetchMode = async (response: Response): Promise<Usage> => {
     const data = (await response.json()) as FetchResponse;
-    if (data.model) {
-      let cost = 0;
-      cost +=
-        (data.usage.prompt_tokens ?? 0) *
-        (models[data.model]?.price?.prompt ?? 0);
-      cost +=
-        (data.usage.completion_tokens ?? 0) *
-        (models[data.model]?.price?.completion ?? 0);
-      chatStore.cost += cost;
-      addTotalCost(cost);
-    }
     const msg = client.processFetchResponse(data);
 
     chatStore.history.push({
@@ -355,12 +344,30 @@ export default function ChatBOX() {
       // calculate cost
       if (usage.response_model_name) {
         let cost = 0;
+
+        if (usage.prompt_tokens_details) {
+          const cached_prompt_tokens =
+            usage.prompt_tokens_details.cached_tokens ?? 0;
+          const uncached_prompt_tokens =
+            usage.prompt_tokens - cached_prompt_tokens;
+          const prompt_price =
+            models[usage.response_model_name]?.price?.prompt ?? 0;
+          const cached_price =
+            models[usage.response_model_name]?.price?.cached_prompt ??
+            prompt_price;
+          cost +=
+            cached_prompt_tokens * cached_price +
+            uncached_prompt_tokens * prompt_price;
+        } else {
           cost +=
             usage.prompt_tokens *
             (models[usage.response_model_name]?.price?.prompt ?? 0);
+        }
+
         cost +=
           usage.completion_tokens *
           (models[usage.response_model_name]?.price?.completion ?? 0);
+
         addTotalCost(cost);
         chatStore.cost += cost;
         console.log("cost", cost);

src/types/models.ts

@@ -7,7 +7,7 @@ interface Model {
   };
 }
 
-const M = 1000 / 1000; // dollars per million tokens
+const M = 1000 * 1000; // dollars per million tokens
 const K = 1000; // dollars per thousand tokens
 
 export const models: Record<string, Model> = {
@@ -37,19 +37,19 @@ export const models: Record<string, Model> = {
   },
   o1: {
     maxToken: 128000,
-    price: { prompt: 15 / M, cached_prompt: 7.5 / M, completion: 60 },
+    price: { prompt: 15 / M, cached_prompt: 7.5 / M, completion: 60 / M },
   },
   "o1-2024-12-17": {
     maxToken: 128000,
-    price: { prompt: 15 / M, cached_prompt: 7.5 / M, completion: 60 },
+    price: { prompt: 15 / M, cached_prompt: 7.5 / M, completion: 60 / M },
   },
   "o1-preview": {
     maxToken: 128000,
-    price: { prompt: 15 / M, cached_prompt: 7.5 / M, completion: 60 },
+    price: { prompt: 15 / M, cached_prompt: 7.5 / M, completion: 60 / M },
   },
   "o1-preview-2024-09-12": {
     maxToken: 128000,
-    price: { prompt: 15 / M, cached_prompt: 7.5 / M, completion: 60 },
+    price: { prompt: 15 / M, cached_prompt: 7.5 / M, completion: 60 / M },
   },
   "o1-mini": {
     maxToken: 128000,
@@ -93,7 +93,7 @@ export const models: Record<string, Model> = {
   },
   "gpt-4-1106-vision-preview": {
     maxToken: 128000,
-    price: { prompt: 0.01 / 1000, completion: 0.03 / 1000 },
+    price: { prompt: 10 / M, completion: 30 / M },
   },
   "gpt-3.5-turbo-0125": {
     maxToken: 16000,