saas: llm usage gets a x3 multiplier (#124)

* pricing-change: llm consumptions x3

* fix: llm usage condition

Author: nomagick

src/tools/read.ts

@@ -58,7 +58,7 @@ export async function readUrl(
     tokenTracker.trackUsage('read', {
       totalTokens: tokens,
       promptTokens: url.length,
-      completionTokens: tokens
+      completionTokens: 0
     });
 
     return { response: data };

src/utils/token-tracker.ts

@@ -31,9 +31,11 @@ export class TokenTracker extends EventEmitter {
 
   getTotalUsage(): LanguageModelUsage {
     return this.usages.reduce((acc, { usage }) => {
-      acc.promptTokens += usage.promptTokens;
-      acc.completionTokens += usage.completionTokens;
-      acc.totalTokens += usage.totalTokens;
+      // CompletionTokens > 0 means LLM usage, apply 3x multiplier
+      const scaler = usage.completionTokens > 0 ? 3 : 1;
+      acc.promptTokens += usage.promptTokens * scaler;
+      acc.completionTokens += usage.completionTokens * scaler;
+      acc.totalTokens += usage.totalTokens * scaler;
       return acc;
     }, { promptTokens: 0, completionTokens: 0, totalTokens: 0 });
   }