@codehz/ai 0.1.7 → 0.2.0

package/dist/index.d.mts

@@ -99,9 +99,7 @@ type Usage = {
   reasoningTokens?: number;
   totalTokens?: number; /** Tokens read from prompt cache (OpenAI cached_tokens, Anthropic cache_read_input_tokens) */
   cachedInputTokens?: number; /** Tokens written to prompt cache (Anthropic cache_creation_input_tokens) */
-  cacheWriteInputTokens?: number; /** Best-effort billable input (full-rate input; excludes discounted cache reads where known) */
-  billableInputTokens?: number; /** Best-effort billable output (non-reasoning slice when provider gives reasoning breakdown) */
-  billableOutputTokens?: number;
+  cacheWriteInputTokens?: number;
 };
 type BillingInfo = {
   amount?: number;
@@ -1084,7 +1082,7 @@ declare function usageFromAnthropicMessages(raw: {
   cache_read_input_tokens?: number;
   [key: string]: unknown;
 }): Partial<Usage>;
-/** Ollama 流式 chunk（无 cache / reasoning 细分时仅填基础与 billable 镜像） */
+/** Ollama 流式 chunk */
 declare function usageFromOllama(raw: {
   prompt_eval_count?: number;
   eval_count?: number;

package/dist/index.mjs

@@ -1020,16 +1020,6 @@ function record(obj) {
 	for (const [key, value] of Object.entries(obj)) if (value !== void 0) out[key] = value;
 	return out;
 }
-function billableFromOpenAIStyle(inputTokens, outputTokens, cachedInputTokens, reasoningTokens) {
-	let billableInputTokens;
-	if (inputTokens !== void 0) billableInputTokens = cachedInputTokens !== void 0 ? Math.max(0, inputTokens - cachedInputTokens) : inputTokens;
-	let billableOutputTokens;
-	if (outputTokens !== void 0) billableOutputTokens = reasoningTokens !== void 0 ? Math.max(0, outputTokens - reasoningTokens) : outputTokens;
-	return record({
-		billableInputTokens,
-		billableOutputTokens
-	});
-}
 /** OpenAI Chat Completions `usage` */
 function usageFromChatCompletions(raw) {
 	const inputTokens = num(raw.prompt_tokens);
@@ -1041,8 +1031,7 @@ function usageFromChatCompletions(raw) {
 		outputTokens,
 		totalTokens: num(raw.total_tokens) ?? (inputTokens !== void 0 && outputTokens !== void 0 ? inputTokens + outputTokens : void 0),
 		cachedInputTokens,
-		reasoningTokens,
-		...billableFromOpenAIStyle(inputTokens, outputTokens, cachedInputTokens, reasoningTokens)
+		reasoningTokens
 	});
 }
 /** OpenAI Responses API `usage` */
@@ -1056,45 +1045,37 @@ function usageFromOpenAIResponses(raw) {
 		outputTokens,
 		totalTokens: num(raw.total_tokens) ?? (inputTokens !== void 0 && outputTokens !== void 0 ? inputTokens + outputTokens : void 0),
 		cachedInputTokens,
-		reasoningTokens,
-		...billableFromOpenAIStyle(inputTokens, outputTokens, cachedInputTokens, reasoningTokens)
+		reasoningTokens
 	});
 }
 /** Anthropic Messages `usage`（message_start / message_delta） */
 function usageFromAnthropicMessages(raw) {
-	const inputTokens = num(raw.input_tokens);
+	const uncachedInputTokens = num(raw.input_tokens);
 	const outputTokens = num(raw.output_tokens);
 	const cacheWriteInputTokens = num(raw.cache_creation_input_tokens);
 	const cachedInputTokens = num(raw.cache_read_input_tokens);
 	const inputParts = [
-		inputTokens,
+		uncachedInputTokens,
 		cacheWriteInputTokens,
 		cachedInputTokens
 	].filter((n) => n !== void 0);
-	const summedInput = inputParts.length > 0 ? inputParts.reduce((sum, n) => sum + n, 0) : void 0;
-	const totalTokens = summedInput !== void 0 && outputTokens !== void 0 ? summedInput + outputTokens : void 0;
-	let billableInputTokens;
-	if (inputTokens !== void 0 || cacheWriteInputTokens !== void 0) billableInputTokens = (inputTokens ?? 0) + (cacheWriteInputTokens ?? 0);
+	const inputTokens = inputParts.length > 0 ? inputParts.reduce((sum, n) => sum + n, 0) : void 0;
 	return record({
 		inputTokens,
 		outputTokens,
-		totalTokens,
+		totalTokens: inputTokens !== void 0 && outputTokens !== void 0 ? inputTokens + outputTokens : void 0,
 		cachedInputTokens,
-		cacheWriteInputTokens,
-		billableInputTokens,
-		billableOutputTokens: outputTokens
+		cacheWriteInputTokens
 	});
 }
-/** Ollama 流式 chunk（无 cache / reasoning 细分时仅填基础与 billable 镜像） */
+/** Ollama 流式 chunk */
 function usageFromOllama(raw) {
 	const inputTokens = num(raw.prompt_eval_count);
 	const outputTokens = num(raw.eval_count);
 	return record({
 		inputTokens,
 		outputTokens,
-		totalTokens: inputTokens !== void 0 && outputTokens !== void 0 ? inputTokens + outputTokens : void 0,
-		billableInputTokens: inputTokens,
-		billableOutputTokens: outputTokens
+		totalTokens: inputTokens !== void 0 && outputTokens !== void 0 ? inputTokens + outputTokens : void 0
 	});
 }
 //#endregion