@thanh01.pmt/curriculum-kit 1.0.8 → 1.0.10

package/dist/media/index.cjs

@@ -427,7 +427,7 @@ function isModelAllowed(model) {
 }
 function getDesignatedFallbackChain() {
   const rawProviders = resolveApiKey2("FALLBACK_AI_PROVIDER") || resolveApiKey2("FALLBACK_AI_PROVIDERS") || resolveApiKey2("DEFAULT_AI_PROVIDER") || "nvidia,openrouter";
-  const rawModels = resolveApiKey2("FALLBACK_AI_MODEL") || resolveApiKey2("FALLBACK_AI_MODELS") || resolveApiKey2("DEFAULT_AI_MODEL") || "nvidia/nemotron-3-ultra-550b-a55b,@preset/coding-free,deepseek/deepseek-chat:free";
+  const rawModels = resolveApiKey2("FALLBACK_AI_MODEL") || resolveApiKey2("FALLBACK_AI_MODELS") || resolveApiKey2("DEFAULT_AI_MODEL") || "nvidia/nemotron-3-ultra-550b-a55b:free,@preset/coding-free,deepseek/deepseek-chat:free";
   const providerList = rawProviders.split(",").map((p) => p.trim().toLowerCase()).filter((p) => p.length > 0 && isProviderEnabled(p));
   const modelList = rawModels.split(",").map((m) => m.trim()).filter((m) => m.length > 0 && isModelAllowed(m));
   const chain = [];
@@ -436,7 +436,7 @@ function getDesignatedFallbackChain() {
     if (provider === "nvidia") {
       providerModels = modelList.filter((m) => m.startsWith("nvidia/") || m.startsWith("meta/") || m.startsWith("mistralai/"));
       if (providerModels.length === 0) {
-        providerModels = ["nvidia/nemotron-3-ultra-550b-a55b", "nvidia/nemotron-3.5-lightning-30b-a3b"];
+        providerModels = ["nvidia/nemotron-3-ultra-550b-a55b:free", "nvidia/nemotron-3.5-lightning-30b-a3b"];
       }
     } else if (provider === "openrouter") {
       providerModels = modelList.filter((m) => m.startsWith("@") || m.includes("/") || m.includes(":"));
@@ -656,6 +656,32 @@ Guidelines:
 3. Keep technical terminology and official API symbols intact.
 4. Format output using clean Markdown, clear sections, bullet points, and code blocks where required.
 5. Do NOT use emojis in headings, tab titles, or UI navigation labels.`;
+  if (options.customInference) {
+    try {
+      const text = await options.customInference({
+        systemInstruction,
+        userPrompt: messages.map((m) => m.content).join("\n\n"),
+        messages,
+        temperature,
+        maxTokens,
+        onChunk: onChunk ?? (() => {
+        })
+      });
+      if (text && text.trim()) return text;
+      throw createAiInferenceError({
+        errorCode: "ERR_AI_ALL_PROVIDERS_FAILED",
+        message: "customInference returned empty output",
+        rawError: { customInference: "empty response" }
+      });
+    } catch (e) {
+      if (e?.errorCode || e?.name === "AiInferenceError") throw e;
+      throw createAiInferenceError({
+        errorCode: "ERR_AI_ALL_PROVIDERS_FAILED",
+        message: `customInference failed: ${e?.message || e}`,
+        rawError: { customInference: String(e?.message || e) }
+      });
+    }
+  }
   const formattedMessages = [
     { role: "system", content: systemInstruction },
     ...messages.map((m) => ({
@@ -668,8 +694,8 @@ Guidelines:
   const isNvidiaPreferred = requestedModel.startsWith("nvidia/") || Boolean(nvidiaKey && !openrouterKey) && isProviderEnabled("nvidia");
   if (nvidiaKey && isProviderEnabled("nvidia")) {
     const candidateNemotronModels = [
-      requestedModel.startsWith("nvidia/") ? requestedModel : "nvidia/nemotron-3-ultra-550b-a55b",
-      "nvidia/nemotron-3-ultra-550b-a55b",
+      requestedModel.startsWith("nvidia/") ? requestedModel : "nvidia/nemotron-3-ultra-550b-a55b:free",
+      "nvidia/nemotron-3-ultra-550b-a55b:free",
       "nvidia/nemotron-3.5-lightning-30b-a3b",
       "nvidia/nemotron-3-super-120b-a12b",
       "nvidia/llama-3.1-nemotron-70b-instruct"