@superatomai/sdk-node 0.0.40 → 0.0.41

package/dist/index.mjs

@@ -3386,30 +3386,47 @@ import { jsonrepair } from "jsonrepair";
 import fs5 from "fs";
 import path4 from "path";
 var PRICING = {
-  // Anthropic
-  "claude-3-5-sonnet-20241022": { input: 3, output: 15, cacheRead: 0.3, cacheWrite: 3.75 },
+  // Anthropic (December 2025)
+  "claude-opus-4-5": { input: 5, output: 25, cacheRead: 0.5, cacheWrite: 6.25 },
+  "claude-opus-4-5-20251101": { input: 5, output: 25, cacheRead: 0.5, cacheWrite: 6.25 },
   "claude-sonnet-4-5": { input: 3, output: 15, cacheRead: 0.3, cacheWrite: 3.75 },
   "claude-sonnet-4-5-20250929": { input: 3, output: 15, cacheRead: 0.3, cacheWrite: 3.75 },
-  "claude-3-5-haiku-20241022": { input: 0.8, output: 4, cacheRead: 0.08, cacheWrite: 1 },
-  "claude-haiku-4-5-20251001": { input: 0.8, output: 4, cacheRead: 0.08, cacheWrite: 1 },
+  "claude-haiku-4-5": { input: 1, output: 5, cacheRead: 0.1, cacheWrite: 1.25 },
+  "claude-haiku-4-5-20251001": { input: 1, output: 5, cacheRead: 0.1, cacheWrite: 1.25 },
+  "claude-3-5-sonnet-20241022": { input: 3, output: 15, cacheRead: 0.3, cacheWrite: 3.75 },
+  "claude-3-5-haiku-20241022": { input: 1, output: 5, cacheRead: 0.1, cacheWrite: 1.25 },
   "claude-3-opus-20240229": { input: 15, output: 75, cacheRead: 1.5, cacheWrite: 18.75 },
   "claude-3-sonnet-20240229": { input: 3, output: 15, cacheRead: 0.3, cacheWrite: 3.75 },
   "claude-3-haiku-20240307": { input: 0.25, output: 1.25, cacheRead: 0.03, cacheWrite: 0.3 },
-  // OpenAI
-  "gpt-4o": { input: 2.5, output: 10 },
+  // OpenAI (December 2025)
+  "gpt-5": { input: 1.25, output: 10 },
+  "gpt-5-mini": { input: 0.25, output: 2 },
+  "gpt-4o": { input: 5, output: 15 },
+  // Updated pricing as of late 2025
   "gpt-4o-mini": { input: 0.15, output: 0.6 },
   "gpt-4-turbo": { input: 10, output: 30 },
   "gpt-4": { input: 30, output: 60 },
   "gpt-3.5-turbo": { input: 0.5, output: 1.5 },
-  // Gemini
+  // Google Gemini (December 2025)
+  "gemini-3-pro": { input: 2, output: 8 },
+  // New Gemini 3
+  "gemini-2.5-pro": { input: 1.25, output: 10 },
+  // For prompts ≤200K tokens, 2x for >200K
+  "gemini-2.5-flash": { input: 0.15, output: 0.6 },
+  // Standard mode (thinking disabled: $0.60, thinking enabled: $3.50)
+  "gemini-2.5-flash-lite": { input: 0.1, output: 0.4 },
+  "gemini-2.0-flash": { input: 0.1, output: 0.4 },
+  "gemini-2.0-flash-lite": { input: 0.075, output: 0.3 },
   "gemini-1.5-pro": { input: 1.25, output: 5 },
   "gemini-1.5-flash": { input: 0.075, output: 0.3 },
-  "gemini-2.0-flash-exp": { input: 0.1, output: 0.4 },
-  // Groq (very cheap)
+  // Groq (December 2025)
   "llama-3.3-70b-versatile": { input: 0.59, output: 0.79 },
   "llama-3.1-70b-versatile": { input: 0.59, output: 0.79 },
   "llama-3.1-8b-instant": { input: 0.05, output: 0.08 },
-  "mixtral-8x7b-32768": { input: 0.24, output: 0.24 }
+  "llama-4-scout-17b-16e": { input: 0.11, output: 0.34 },
+  "llama-4-maverick-17b-128e": { input: 0.2, output: 0.6 },
+  "mixtral-8x7b-32768": { input: 0.27, output: 0.27 },
+  "qwen3-32b": { input: 0.29, output: 0.59 }
 };
 var DEFAULT_PRICING = { input: 3, output: 15 };
 var LLMUsageLogger = class {
@@ -5336,6 +5353,38 @@ var BaseLLM = class {
     this.fastModel = config?.fastModel || this.getDefaultFastModel();
     this.defaultLimit = config?.defaultLimit || 50;
     this.apiKey = config?.apiKey;
+    this.modelStrategy = config?.modelStrategy || "fast";
+  }
+  /**
+   * Get the appropriate model based on task type and model strategy
+   * @param taskType - 'complex' for text generation/matching, 'simple' for classification/actions
+   * @returns The model string to use for this task
+   */
+  getModelForTask(taskType) {
+    switch (this.modelStrategy) {
+      case "best":
+        return this.model;
+      case "fast":
+        return this.fastModel;
+      case "balanced":
+      default:
+        return taskType === "complex" ? this.model : this.fastModel;
+    }
+  }
+  /**
+   * Set the model strategy at runtime
+   * @param strategy - 'best', 'fast', or 'balanced'
+   */
+  setModelStrategy(strategy) {
+    this.modelStrategy = strategy;
+    logger.info(`[${this.getProviderName()}] Model strategy set to: ${strategy}`);
+  }
+  /**
+   * Get the current model strategy
+   * @returns The current model strategy
+   */
+  getModelStrategy() {
+    return this.modelStrategy;
   }
   /**
    * Get the API key (from instance, parameter, or environment)
@@ -5520,7 +5569,7 @@ ${JSON.stringify(tool.requiredFields || [], null, 2)}`;
           user: prompts.user
         },
         {
-          model: this.model,
+          model: this.getModelForTask("complex"),
           maxTokens: 8192,
           temperature: 0.2,
           apiKey: this.getApiKey(apiKey),
@@ -5643,7 +5692,7 @@ ${JSON.stringify(tool.requiredFields || [], null, 2)}`;
           user: prompts.user
         },
         {
-          model: this.fastModel,
+          model: this.getModelForTask("simple"),
           maxTokens: 1500,
           temperature: 0.2,
           apiKey: this.getApiKey(apiKey)
@@ -5704,7 +5753,7 @@ ${JSON.stringify(tool.requiredFields || [], null, 2)}`;
           user: prompts.user
         },
         {
-          model: this.model,
+          model: this.getModelForTask("complex"),
           maxTokens: 3e3,
           temperature: 0.2,
           apiKey: this.getApiKey(apiKey)
@@ -6194,7 +6243,7 @@ ${errorMsg}
         tools,
         toolHandler,
         {
-          model: this.model,
+          model: this.getModelForTask("complex"),
           maxTokens: 4e3,
           temperature: 0.7,
           apiKey: this.getApiKey(apiKey),
@@ -6239,6 +6288,21 @@ ${errorMsg}
       if (category === "general") {
         logger.info(`[${this.getProviderName()}] Skipping component generation for general/conversational question`);
         logCollector?.info("Skipping component generation for general question");
+        logger.info(`[${this.getProviderName()}] Generating actions for general question...`);
+        const nextQuestions = await this.generateNextQuestions(
+          userPrompt,
+          null,
+          // no component
+          void 0,
+          // no component data
+          apiKey,
+          logCollector,
+          conversationHistory,
+          textResponse
+          // pass text response as context
+        );
+        actions = convertQuestionsToActions(nextQuestions);
+        logger.info(`[${this.getProviderName()}] Generated ${actions.length} follow-up actions for general question`);
       } else if (components && components.length > 0) {
         logger.info(`[${this.getProviderName()}] Matching components from text response...`);
         const componentStreamCallback = wrappedStreamCallback && category !== "data_modification" ? (component) => {
@@ -6397,10 +6461,18 @@ ${errorMsg}
             logger.info(`[${this.getProviderName()}] \u2713 100% match - returning UI block directly without adaptation`);
             logCollector?.info(`\u2713 Exact match (${(conversationMatch.similarity * 100).toFixed(2)}%) - returning cached result`);
             logCollector?.info(`Total time taken: ${elapsedTime2}ms (${(elapsedTime2 / 1e3).toFixed(2)}s)`);
+            if (streamCallback && cachedTextResponse) {
+              logger.info(`[${this.getProviderName()}] Streaming cached text response to frontend`);
+              streamCallback(cachedTextResponse);
+            }
+            const cachedActions = conversationMatch.uiBlock?.actions || [];
             return {
               success: true,
               data: {
+                text: cachedTextResponse,
                 component,
+                matchedComponents: component?.props?.config?.components || [],
+                actions: cachedActions,
                 reasoning: `Exact match from previous conversation (${(conversationMatch.similarity * 100).toFixed(2)}% similarity)`,
                 method: `${this.getProviderName()}-semantic-match-exact`,
                 semanticSimilarity: conversationMatch.similarity
@@ -6423,10 +6495,18 @@ ${errorMsg}
             logger.info(`[${this.getProviderName()}] Total time taken: ${elapsedTime2}ms (${(elapsedTime2 / 1e3).toFixed(2)}s)`);
             logCollector?.info(`\u2713 UI block adapted successfully`);
             logCollector?.info(`Total time taken: ${elapsedTime2}ms (${(elapsedTime2 / 1e3).toFixed(2)}s)`);
+            if (streamCallback && cachedTextResponse) {
+              logger.info(`[${this.getProviderName()}] Streaming cached text response to frontend (adapted match)`);
+              streamCallback(cachedTextResponse);
+            }
+            const cachedActions = conversationMatch.uiBlock?.actions || [];
             return {
               success: true,
               data: {
+                text: cachedTextResponse,
                 component: adaptResult.adaptedComponent,
+                matchedComponents: adaptResult.adaptedComponent?.props?.config?.components || [],
+                actions: cachedActions,
                 reasoning: `Adapted from previous conversation: ${originalPrompt}`,
                 method: `${this.getProviderName()}-semantic-match`,
                 semanticSimilarity: conversationMatch.similarity,
@@ -6539,15 +6619,26 @@ ${errorMsg}
   /**
    * Generate next questions that the user might ask based on the original prompt and generated component
    * This helps provide intelligent suggestions for follow-up queries
+   * For general/conversational questions without components, pass textResponse instead
    */
-  async generateNextQuestions(originalUserPrompt, component, componentData, apiKey, logCollector, conversationHistory) {
+  async generateNextQuestions(originalUserPrompt, component, componentData, apiKey, logCollector, conversationHistory, textResponse) {
     try {
-      const component_info = `
+      let component_info;
+      if (component) {
+        component_info = `
 				Component Name: ${component.name}
 				Component Type: ${component.type}
 				Component Description: ${component.description || "No description"}
 				Component Props: ${component.props ? JSON.stringify(component.props, null, 2) : "No props"}
 			`;
+      } else if (textResponse) {
+        component_info = `
+				Response Type: Text/Conversational Response
+				Response Content: ${textResponse.substring(0, 1e3)}${textResponse.length > 1e3 ? "..." : ""}
+			`;
+      } else {
+        component_info = "No component or response context available";
+      }
       const component_data = componentData ? `Component Data: ${JSON.stringify(componentData, null, 2)}` : "";
       const prompts = await promptLoader.loadPrompts("actions", {
         ORIGINAL_USER_PROMPT: originalUserPrompt,
@@ -6561,7 +6652,7 @@ ${errorMsg}
           user: prompts.user
         },
         {
-          model: this.fastModel,
+          model: this.getModelForTask("simple"),
           maxTokens: 1200,
           temperature: 0.7,
           apiKey: this.getApiKey(apiKey)
@@ -11088,7 +11179,9 @@ var SuperatomSDK = class {
     this.openaiApiKey = config.OPENAI_API_KEY || process.env.OPENAI_API_KEY || "";
     this.llmProviders = config.LLM_PROVIDERS || getLLMProviders();
     this.databaseType = config.databaseType || "postgresql";
-    logger.info(`Initializing Superatom SDK v${SDK_VERSION} for project ${this.projectId}, llm providers: ${this.llmProviders.join(", ")}, database type: ${this.databaseType}`);
+    this.modelStrategy = config.modelStrategy || "fast";
+    this.applyModelStrategy(this.modelStrategy);
+    logger.info(`Initializing Superatom SDK v${SDK_VERSION} for project ${this.projectId}, llm providers: ${this.llmProviders.join(", ")}, database type: ${this.databaseType}, model strategy: ${this.modelStrategy}`);
     this.userManager = new UserManager(this.projectId, 5e3);
     this.dashboardManager = new DashboardManager(this.projectId);
     this.reportManager = new ReportManager(this.projectId);
@@ -11467,6 +11560,31 @@ var SuperatomSDK = class {
   getTools() {
     return this.tools;
   }
+  /**
+   * Apply model strategy to all LLM provider singletons
+   * @param strategy - 'best', 'fast', or 'balanced'
+   */
+  applyModelStrategy(strategy) {
+    anthropicLLM.setModelStrategy(strategy);
+    groqLLM.setModelStrategy(strategy);
+    geminiLLM.setModelStrategy(strategy);
+    openaiLLM.setModelStrategy(strategy);
+    logger.info(`Model strategy '${strategy}' applied to all LLM providers`);
+  }
+  /**
+   * Set model strategy at runtime
+   * @param strategy - 'best', 'fast', or 'balanced'
+   */
+  setModelStrategy(strategy) {
+    this.modelStrategy = strategy;
+    this.applyModelStrategy(strategy);
+  }
+  /**
+   * Get current model strategy
+   */
+  getModelStrategy() {
+    return this.modelStrategy;
+  }
 };
 export {
   BM25L,
@@ -11481,9 +11599,13 @@ export {
   UIBlock,
   UILogCollector,
   UserManager,
+  anthropicLLM,
+  geminiLLM,
+  groqLLM,
   hybridRerank,
   llmUsageLogger,
   logger,
+  openaiLLM,
   rerankChromaResults,
   rerankConversationResults,
   userPromptErrorLogger