vibezcheck 0.5.3 → 0.5.4

package/dist/ai-sdk/index.js

@@ -309,12 +309,38 @@ function extractOpenAIResponseUsage(response) {
 function inspectOpenAIStreamChunk(chunk) {
   if (!chunk || typeof chunk !== "object") return {};
   const model = chunk.model;
-  const rawUsage = chunk.usage || chunk.token_usage || chunk.x_groq?.usage || chunk.usageMetadata;
+  const rawUsage = (chunk.type === "finish" ? chunk.usage : null) || chunk.usage || chunk.token_usage || chunk.x_groq?.usage || chunk.usageMetadata;
   if (rawUsage) {
-    const inputTokens = rawUsage.prompt_tokens ?? rawUsage.input_tokens ?? rawUsage.promptTokens ?? rawUsage.inputTokens ?? rawUsage.promptTokenCount ?? 0;
-    const outputTokens = rawUsage.completion_tokens ?? rawUsage.output_tokens ?? rawUsage.completionTokens ?? rawUsage.outputTokens ?? rawUsage.candidatesTokenCount ?? 0;
-    const reasoningTokens = rawUsage.completion_tokens_details?.reasoning_tokens ?? rawUsage.output_token_details?.reasoning_tokens ?? rawUsage.reasoning_tokens ?? rawUsage.reasoningTokens ?? rawUsage.thoughtsTokenCount ?? 0;
-    const cachedTokens = rawUsage.prompt_tokens_details?.cached_tokens ?? rawUsage.input_token_details?.cached_tokens ?? rawUsage.prompt_cache_hit_tokens ?? rawUsage.cached_tokens ?? rawUsage.cachedTokens ?? rawUsage.cachedContentTokenCount ?? 0;
+    let inputTokens = 0;
+    let outputTokens = 0;
+    let reasoningTokens = 0;
+    let cachedTokens = 0;
+    if (typeof rawUsage.inputTokens === "object" && rawUsage.inputTokens !== null) {
+      inputTokens = rawUsage.inputTokens.total ?? rawUsage.inputTokens.noCache ?? 0;
+      cachedTokens = rawUsage.inputTokens.cacheRead ?? 0;
+    } else if (typeof rawUsage.promptTokens === "number") {
+      inputTokens = rawUsage.promptTokens;
+    } else if (typeof rawUsage.inputTokens === "number") {
+      inputTokens = rawUsage.inputTokens;
+    } else {
+      inputTokens = rawUsage.prompt_tokens ?? rawUsage.input_tokens ?? rawUsage.promptTokenCount ?? 0;
+    }
+    if (typeof rawUsage.outputTokens === "object" && rawUsage.outputTokens !== null) {
+      outputTokens = rawUsage.outputTokens.total ?? rawUsage.outputTokens.text ?? 0;
+      reasoningTokens = rawUsage.outputTokens.reasoning ?? 0;
+    } else if (typeof rawUsage.completionTokens === "number") {
+      outputTokens = rawUsage.completionTokens;
+    } else if (typeof rawUsage.outputTokens === "number") {
+      outputTokens = rawUsage.outputTokens;
+    } else {
+      outputTokens = rawUsage.completion_tokens ?? rawUsage.output_tokens ?? rawUsage.candidatesTokenCount ?? 0;
+    }
+    if (!reasoningTokens) {
+      reasoningTokens = rawUsage.completion_tokens_details?.reasoning_tokens ?? rawUsage.output_token_details?.reasoning_tokens ?? rawUsage.reasoning_tokens ?? rawUsage.reasoningTokens ?? rawUsage.thoughtsTokenCount ?? 0;
+    }
+    if (!cachedTokens) {
+      cachedTokens = rawUsage.prompt_tokens_details?.cached_tokens ?? rawUsage.input_token_details?.cached_tokens ?? rawUsage.prompt_cache_hit_tokens ?? rawUsage.cached_tokens ?? rawUsage.cachedTokens ?? rawUsage.cachedContentTokenCount ?? 0;
+    }
     return {
       model,
       usage: {
@@ -460,10 +486,36 @@ function extractGenericResponseUsage(response, fallbackModel = "generic-llm", fa
   if (!response || typeof response !== "object") return null;
   const usage = response.usage || response.token_usage || response.usageMetadata;
   if (usage) {
-    const inputTokens = usage.prompt_tokens ?? usage.input_tokens ?? usage.promptTokenCount ?? usage.prompt_eval_count ?? 0;
-    const outputTokens = usage.completion_tokens ?? usage.output_tokens ?? usage.candidatesTokenCount ?? usage.eval_count ?? 0;
-    const reasoningTokens = usage.reasoning_tokens ?? usage.thoughtsTokenCount ?? usage.completion_tokens_details?.reasoning_tokens ?? 0;
-    const cachedTokens = usage.prompt_tokens_details?.cached_tokens ?? usage.cached_tokens ?? usage.cachedContentTokenCount ?? 0;
+    let inputTokens = 0;
+    let outputTokens = 0;
+    let reasoningTokens = 0;
+    let cachedTokens = 0;
+    if (typeof usage.inputTokens === "object" && usage.inputTokens !== null) {
+      inputTokens = usage.inputTokens.total ?? usage.inputTokens.noCache ?? 0;
+      cachedTokens = usage.inputTokens.cacheRead ?? 0;
+    } else if (typeof usage.promptTokens === "number") {
+      inputTokens = usage.promptTokens;
+    } else if (typeof usage.inputTokens === "number") {
+      inputTokens = usage.inputTokens;
+    } else {
+      inputTokens = usage.prompt_tokens ?? usage.input_tokens ?? usage.promptTokenCount ?? usage.prompt_eval_count ?? 0;
+    }
+    if (typeof usage.outputTokens === "object" && usage.outputTokens !== null) {
+      outputTokens = usage.outputTokens.total ?? usage.outputTokens.text ?? 0;
+      reasoningTokens = usage.outputTokens.reasoning ?? 0;
+    } else if (typeof usage.completionTokens === "number") {
+      outputTokens = usage.completionTokens;
+    } else if (typeof usage.outputTokens === "number") {
+      outputTokens = usage.outputTokens;
+    } else {
+      outputTokens = usage.completion_tokens ?? usage.output_tokens ?? usage.candidatesTokenCount ?? usage.eval_count ?? 0;
+    }
+    if (!reasoningTokens) {
+      reasoningTokens = usage.reasoning_tokens ?? usage.thoughtsTokenCount ?? usage.completion_tokens_details?.reasoning_tokens ?? 0;
+    }
+    if (!cachedTokens) {
+      cachedTokens = usage.prompt_tokens_details?.cached_tokens ?? usage.cached_tokens ?? usage.cachedContentTokenCount ?? 0;
+    }
     const model = response.model || fallbackModel;
     const provider = response.provider || fallbackProvider;
     return {
@@ -515,6 +567,10 @@ var MODEL_PRICING_TABLE = {
   "chatgpt-4o-latest": { inputPer1M: 5, outputPer1M: 15 },
   "gpt-4.1": { inputPer1M: 2, outputPer1M: 8, cachedInputPer1M: 1 },
   "gpt-4.1-nano": { inputPer1M: 0.1, outputPer1M: 0.4, cachedInputPer1M: 0.05 },
+  "gpt-6-astra": { inputPer1M: 5, outputPer1M: 25, cachedInputPer1M: 0.5 },
+  "gpt-6-astra-fast": { inputPer1M: 3, outputPer1M: 15, cachedInputPer1M: 0.3 },
+  "gpt-image-2.5-flare": { inputPer1M: 0, outputPer1M: 0.04 },
+  "gpt-image-2.5-sunburst": { inputPer1M: 0, outputPer1M: 0.08 },
   // --- OpenAI (Legacy & Backward Compatibility) ---
   "gpt-4-turbo": { inputPer1M: 10, outputPer1M: 30, cachedInputPer1M: 5 },
   "gpt-4-turbo-preview": { inputPer1M: 10, outputPer1M: 30, cachedInputPer1M: 5 },
@@ -567,7 +623,20 @@ var MODEL_PRICING_TABLE = {
   "mistral-large-2411": { inputPer1M: 2, outputPer1M: 6 },
   "codestral-latest": { inputPer1M: 0.3, outputPer1M: 0.9 },
   "mistral-medium-latest": { inputPer1M: 2.7, outputPer1M: 8.1 },
+  "mistral-medium-3-5": { inputPer1M: 2, outputPer1M: 6, cachedInputPer1M: 0.4 },
+  "mistral-medium-2604": { inputPer1M: 2, outputPer1M: 6, cachedInputPer1M: 0.4 },
+  "magistral-small-latest": { inputPer1M: 0.5, outputPer1M: 1.5, cachedInputPer1M: 0.1 },
+  "magistral-medium-latest": { inputPer1M: 2, outputPer1M: 6, cachedInputPer1M: 0.4 },
   "mistral-small-latest": { inputPer1M: 0.2, outputPer1M: 0.6 },
+  "mistral-small-2603": { inputPer1M: 0.2, outputPer1M: 0.6, cachedInputPer1M: 0.05 },
+  "mistral-vibe-cli-fast": { inputPer1M: 0.2, outputPer1M: 0.6, cachedInputPer1M: 0.05 },
+  "mistral-vibe-cli-latest": { inputPer1M: 0.2, outputPer1M: 0.6, cachedInputPer1M: 0.05 },
+  "codestral-2508": { inputPer1M: 0.3, outputPer1M: 0.9, cachedInputPer1M: 0.08 },
+  "codestral-embed": { inputPer1M: 0.1, outputPer1M: 0 },
+  "mistral-embed-2312": { inputPer1M: 0.1, outputPer1M: 0 },
+  "zai-glm-5-2": { inputPer1M: 0.6, outputPer1M: 2.2, cachedInputPer1M: 0.15 },
+  "glm-5-2": { inputPer1M: 0.6, outputPer1M: 2.2, cachedInputPer1M: 0.15 },
+  "labs-leanstral-1-5": { inputPer1M: 0.2, outputPer1M: 0.6, cachedInputPer1M: 0.05 },
   "ministral-8b-latest": { inputPer1M: 0.1, outputPer1M: 0.1 },
   "ministral-3b-latest": { inputPer1M: 0.04, outputPer1M: 0.04 },
   "open-mistral-7b": { inputPer1M: 0.2, outputPer1M: 0.2 },
@@ -590,11 +659,17 @@ var MODEL_PRICING_TABLE = {
   "gemma2-9b-it": { inputPer1M: 0.2, outputPer1M: 0.2 },
   // --- DeepSeek ---
   "deepseek-v4-pro": { inputPer1M: 0.66, outputPer1M: 1.98, cachedInputPer1M: 0.15 },
+  "deepseek-v4.1-flash": { inputPer1M: 0.2, outputPer1M: 0.6, cachedInputPer1M: 0.05 },
   "deepseek-v4-flash": { inputPer1M: 0.22, outputPer1M: 0.66, cachedInputPer1M: 0.05 },
   "deepseek-v3": { inputPer1M: 0.14, outputPer1M: 0.28, cachedInputPer1M: 0.014 },
   "deepseek-chat": { inputPer1M: 0.14, outputPer1M: 0.28, cachedInputPer1M: 0.014 },
   "deepseek-r1": { inputPer1M: 0.55, outputPer1M: 2.19, cachedInputPer1M: 0.14 },
   "deepseek-reasoner": { inputPer1M: 0.55, outputPer1M: 2.19, cachedInputPer1M: 0.14 },
+  // --- Gateway, Inception & InclusionAI Models ---
+  "mercury-2.5": { inputPer1M: 0.25, outputPer1M: 0.75, cachedInputPer1M: 0.05 },
+  "mercury-2": { inputPer1M: 0.25, outputPer1M: 0.75, cachedInputPer1M: 0.05 },
+  "ling-3.0-flash-sante": { inputPer1M: 0.15, outputPer1M: 0.45, cachedInputPer1M: 0.03 },
+  "ling-3.0-flash": { inputPer1M: 0.15, outputPer1M: 0.45, cachedInputPer1M: 0.03 },
   // --- Cohere ---
   "command-r-plus": { inputPer1M: 2.5, outputPer1M: 10 },
   "command-r": { inputPer1M: 0.15, outputPer1M: 0.6 },
@@ -642,7 +717,15 @@ var MODEL_ALIASES = {
   // Mistral Shorthands
   "codestral": "codestral-latest",
   "mistral-large": "mistral-large-latest",
-  "mistral-small": "mistral-small-latest"
+  "mistral-small": "mistral-small-latest",
+  "magistral": "magistral-small-latest",
+  "magistral-small": "magistral-small-latest",
+  "magistral-medium": "magistral-medium-latest",
+  "mistral-3.5": "mistral-medium-3-5",
+  "mistral-medium-3.5": "mistral-medium-3-5",
+  "gpt-image-2.5": "gpt-image-2.5-flare",
+  "gpt-image-sunburst": "gpt-image-2.5-sunburst",
+  "gpt-image-flare": "gpt-image-2.5-flare"
 };
 var customPricingRegistry = {};
 function normalizeModelKey(rawModel) {
@@ -1141,11 +1224,41 @@ function withBilling(model, optionsOrCustomer, extraOptions = {}) {
   };
   const handleUsage = (rawUsage, extraMeta) => {
     if (!rawUsage) return;
-    const inputTokens = rawUsage.promptTokens ?? rawUsage.inputTokens ?? rawUsage.prompt_tokens ?? rawUsage.input_tokens ?? rawUsage.promptTokenCount ?? 0;
-    const outputTokens = rawUsage.completionTokens ?? rawUsage.outputTokens ?? rawUsage.completion_tokens ?? rawUsage.output_tokens ?? rawUsage.candidatesTokenCount ?? 0;
-    const reasoningTokens = rawUsage.reasoningTokens ?? rawUsage.completionTokensDetails?.reasoningTokens ?? rawUsage.outputTokenDetails?.reasoningTokens ?? rawUsage.reasoning_tokens ?? rawUsage.completion_tokens_details?.reasoning_tokens ?? rawUsage.output_token_details?.reasoning_tokens ?? rawUsage.thoughtsTokenCount ?? 0;
-    const cachedTokens = rawUsage.promptTokensDetails?.cachedTokens ?? rawUsage.inputTokenDetails?.cachedTokens ?? rawUsage.cachedTokens ?? rawUsage.cached_tokens ?? rawUsage.prompt_tokens_details?.cached_tokens ?? rawUsage.input_token_details?.cached_tokens ?? rawUsage.cachedContentTokenCount ?? rawUsage.cache_read_input_tokens ?? 0;
-    const cacheWriteTokens = rawUsage.cacheWriteTokens ?? rawUsage.cache_creation_input_tokens ?? rawUsage.cacheCreationTokens ?? 0;
+    let inputTokens = 0;
+    let outputTokens = 0;
+    let reasoningTokens = 0;
+    let cachedTokens = 0;
+    let cacheWriteTokens = 0;
+    if (typeof rawUsage.inputTokens === "object" && rawUsage.inputTokens !== null) {
+      inputTokens = rawUsage.inputTokens.total ?? rawUsage.inputTokens.noCache ?? 0;
+      cachedTokens = rawUsage.inputTokens.cacheRead ?? 0;
+      cacheWriteTokens = rawUsage.inputTokens.cacheWrite ?? 0;
+    } else if (typeof rawUsage.promptTokens === "number") {
+      inputTokens = rawUsage.promptTokens;
+    } else if (typeof rawUsage.inputTokens === "number") {
+      inputTokens = rawUsage.inputTokens;
+    } else {
+      inputTokens = rawUsage.prompt_tokens ?? rawUsage.input_tokens ?? rawUsage.promptTokenCount ?? 0;
+    }
+    if (typeof rawUsage.outputTokens === "object" && rawUsage.outputTokens !== null) {
+      outputTokens = rawUsage.outputTokens.total ?? rawUsage.outputTokens.text ?? 0;
+      reasoningTokens = rawUsage.outputTokens.reasoning ?? 0;
+    } else if (typeof rawUsage.completionTokens === "number") {
+      outputTokens = rawUsage.completionTokens;
+    } else if (typeof rawUsage.outputTokens === "number") {
+      outputTokens = rawUsage.outputTokens;
+    } else {
+      outputTokens = rawUsage.completion_tokens ?? rawUsage.output_tokens ?? rawUsage.candidatesTokenCount ?? 0;
+    }
+    if (!reasoningTokens) {
+      reasoningTokens = rawUsage.reasoningTokens ?? rawUsage.completionTokensDetails?.reasoningTokens ?? rawUsage.outputTokenDetails?.reasoningTokens ?? rawUsage.reasoning_tokens ?? rawUsage.completion_tokens_details?.reasoning_tokens ?? rawUsage.output_token_details?.reasoning_tokens ?? rawUsage.thoughtsTokenCount ?? 0;
+    }
+    if (!cachedTokens) {
+      cachedTokens = rawUsage.promptTokensDetails?.cachedTokens ?? rawUsage.inputTokenDetails?.cachedTokens ?? rawUsage.cachedTokens ?? rawUsage.cached_tokens ?? rawUsage.prompt_tokens_details?.cached_tokens ?? rawUsage.input_token_details?.cached_tokens ?? rawUsage.cachedContentTokenCount ?? rawUsage.cache_read_input_tokens ?? 0;
+    }
+    if (!cacheWriteTokens) {
+      cacheWriteTokens = rawUsage.cacheWriteTokens ?? rawUsage.cache_creation_input_tokens ?? rawUsage.cacheCreationTokens ?? 0;
+    }
     const usage = {
       inputTokens,
       outputTokens,
@@ -1346,16 +1459,22 @@ function withBilling(model, optionsOrCustomer, extraOptions = {}) {
           const transformStream = new TransformStream({
             transform(chunk, controller) {
               controller.enqueue(chunk);
-              const delta = chunk.textDelta || chunk.text || chunk.delta;
-              if (delta && typeof delta === "string") {
-                accumulatedChars += delta.length;
-              } else if (chunk.type === "reasoning" || chunk.type === "reasoning-delta") {
-                const reasoning = chunk.reasoning || chunk.textDelta || chunk.text || chunk.delta || "";
-                if (typeof reasoning === "string") {
-                  accumulatedChars += reasoning.length;
+              if (chunk.type === "text-delta" && typeof chunk.delta === "string") {
+                accumulatedChars += chunk.delta.length;
+              } else if (chunk.type === "reasoning-delta" && typeof chunk.delta === "string") {
+                accumulatedChars += chunk.delta.length;
+              } else {
+                const delta = chunk.textDelta || chunk.text || chunk.delta;
+                if (delta && typeof delta === "string") {
+                  accumulatedChars += delta.length;
+                } else if (chunk.type === "reasoning" || chunk.type === "reasoning-delta") {
+                  const reasoning = chunk.reasoning || chunk.textDelta || chunk.text || chunk.delta || "";
+                  if (typeof reasoning === "string") {
+                    accumulatedChars += reasoning.length;
+                  }
                 }
               }
-              const chunkUsage = chunk.usage || chunk.token_usage || chunk.usageMetadata;
+              const chunkUsage = (chunk.type === "finish" ? chunk.usage : null) || chunk.usage || chunk.token_usage || chunk.usageMetadata;
               if (chunkUsage) {
                 streamCompleted = true;
                 handleUsage(chunkUsage);
@@ -1434,12 +1553,16 @@ function createVibezModel(modelOrId, options = {}) {
       providerName = "deepseek";
     } else if (rawId.startsWith("grok-")) {
       providerName = "xai";
-    } else if (rawId.startsWith("mistral-") || rawId.startsWith("codestral-") || rawId.startsWith("ministral-") || rawId.startsWith("open-mistral") || rawId.startsWith("open-mixtral")) {
+    } else if (rawId.startsWith("mistral-") || rawId.startsWith("codestral-") || rawId.startsWith("ministral-") || rawId.startsWith("magistral-") || rawId.startsWith("voxtral-") || rawId.startsWith("devstral-") || rawId.startsWith("open-mistral") || rawId.startsWith("open-mixtral")) {
       providerName = "mistral";
     } else if (rawId.startsWith("llama-") || rawId.startsWith("mixtral-") || rawId.startsWith("gemma") || rawId.startsWith("qwen-")) {
       providerName = "groq";
     } else if (rawId.startsWith("command-")) {
       providerName = "cohere";
+    } else if (rawId.startsWith("mercury-") || rawId.startsWith("ling-")) {
+      providerName = "gateway";
+    } else if (rawId.startsWith("gpt-image-") || rawId.startsWith("gpt-6-")) {
+      providerName = "openai";
     }
     const apiKey = options.apiKey || process.env.AI_GATEWAY_API_KEY || (providerName === "anthropic" ? process.env.ANTHROPIC_API_KEY : void 0) || (providerName === "google" ? process.env.GOOGLE_GENERATIVE_AI_API_KEY : void 0) || (providerName === "mistral" ? process.env.MISTRAL_API_KEY : void 0) || (providerName === "groq" ? process.env.GROQ_API_KEY : void 0) || (providerName === "deepseek" ? process.env.DEEPSEEK_API_KEY : void 0) || (providerName === "xai" ? process.env.XAI_API_KEY : void 0) || (providerName === "cohere" ? process.env.COHERE_API_KEY : void 0) || process.env.OPENAI_API_KEY;
     const baseURL = options.baseURL || process.env.AI_GATEWAY_BASE_URL || (providerName === "openai" ? "https://api.openai.com/v1" : void 0);