gitlab-ai-provider 6.7.0 → 6.7.1

package/dist/index.mjs

@@ -403,14 +403,15 @@ ${message.content}` : message.content;
     return { unified, raw: stopReason ?? void 0 };
   }
   createUsage(params) {
-    const inputTotal = params?.inputTotal;
+    const inputNoCache = params?.inputTotal;
     const outputTotal = params?.outputTotal;
     const cacheRead = params?.cacheRead;
     const cacheWrite = params?.cacheWrite;
+    const inputTotal = inputNoCache != null ? inputNoCache + (cacheRead ?? 0) + (cacheWrite ?? 0) : void 0;
     return {
       inputTokens: {
         total: inputTotal,
-        noCache: inputTotal == null ? void 0 : Math.max(0, inputTotal - (cacheRead ?? 0) - (cacheWrite ?? 0)),
+        noCache: inputNoCache,
         cacheRead,
         cacheWrite
       },
@@ -538,12 +539,13 @@ ${message.content}` : message.content;
                 switch (event.type) {
                   case "message_start":
                     if (event.message.usage) {
+                      const msgUsage = event.message.usage;
                       usage = self.createUsage({
-                        inputTotal: event.message.usage.input_tokens,
+                        inputTotal: msgUsage.input_tokens,
                         outputTotal: usage.outputTokens.total,
                         outputReasoning: usage.outputTokens.reasoning,
-                        cacheRead: usage.inputTokens.cacheRead,
-                        cacheWrite: usage.inputTokens.cacheWrite,
+                        cacheRead: msgUsage.cache_read_input_tokens,
+                        cacheWrite: msgUsage.cache_creation_input_tokens,
                         raw: usage.raw
                       });
                     }
@@ -617,12 +619,13 @@ ${message.content}` : message.content;
                   }
                   case "message_delta":
                     if (event.usage) {
+                      const deltaUsage = event.usage;
                       usage = self.createUsage({
-                        inputTotal: usage.inputTokens.total,
-                        outputTotal: event.usage.output_tokens,
+                        inputTotal: usage.inputTokens.noCache,
+                        outputTotal: deltaUsage.output_tokens,
                         outputReasoning: usage.outputTokens.reasoning,
-                        cacheRead: usage.inputTokens.cacheRead,
-                        cacheWrite: usage.inputTokens.cacheWrite,
+                        cacheRead: deltaUsage.cache_read_input_tokens ?? usage.inputTokens.cacheRead,
+                        cacheWrite: deltaUsage.cache_creation_input_tokens ?? usage.inputTokens.cacheWrite,
                         raw: usage.raw
                       });
                     }
@@ -1610,7 +1613,7 @@ var GitLabOpenAILanguageModel = class {
 import WebSocket from "isomorphic-ws";
 
 // src/version.ts
-var VERSION = true ? "6.6.0" : "0.0.0-dev";
+var VERSION = true ? "6.7.0" : "0.0.0-dev";
 
 // src/gitlab-workflow-types.ts
 var WorkflowType = /* @__PURE__ */ ((WorkflowType2) => {