@aituber-onair/chat 0.59.0 → 0.60.0

package/dist/umd/aituber-onair-chat.js

@@ -82,8 +82,10 @@ var AITuberOnAirChat = (() => {
     MODEL_ANTHROPIC_CLAUDE_OPUS_4: () => MODEL_ANTHROPIC_CLAUDE_OPUS_4,
     MODEL_ANTHROPIC_CLAUDE_OPUS_4_8: () => MODEL_ANTHROPIC_CLAUDE_OPUS_4_8,
     MODEL_ANTHROPIC_CLAUDE_OPUS_5: () => MODEL_ANTHROPIC_CLAUDE_OPUS_5,
+    MODEL_ANTHROPIC_CLAUDE_OPUS_5_5: () => MODEL_ANTHROPIC_CLAUDE_OPUS_5_5,
     MODEL_ANTHROPIC_CLAUDE_SONNET_4: () => MODEL_ANTHROPIC_CLAUDE_SONNET_4,
     MODEL_ANTHROPIC_CLAUDE_SONNET_5: () => MODEL_ANTHROPIC_CLAUDE_SONNET_5,
+    MODEL_ANTHROPIC_CLAUDE_SONNET_5_5: () => MODEL_ANTHROPIC_CLAUDE_SONNET_5_5,
     MODEL_ANTHROPIC_CLAUDE_SONNET_LATEST: () => MODEL_ANTHROPIC_CLAUDE_SONNET_LATEST,
     MODEL_CLAUDE_3_5_HAIKU: () => MODEL_CLAUDE_3_5_HAIKU,
     MODEL_CLAUDE_3_5_SONNET: () => MODEL_CLAUDE_3_5_SONNET,
@@ -144,6 +146,7 @@ var AITuberOnAirChat = (() => {
     MODEL_GLM_5_2: () => MODEL_GLM_5_2,
     MODEL_GLM_5_3: () => MODEL_GLM_5_3,
     MODEL_GLM_5_3_FLASH: () => MODEL_GLM_5_3_FLASH,
+    MODEL_GLM_5_3_FLASHX: () => MODEL_GLM_5_3_FLASHX,
     MODEL_GLM_5_TURBO: () => MODEL_GLM_5_TURBO,
     MODEL_GOOGLE_GEMINI_2_5_FLASH: () => MODEL_GOOGLE_GEMINI_2_5_FLASH,
     MODEL_GOOGLE_GEMINI_2_5_FLASH_LITE_PREVIEW_09_2025: () => MODEL_GOOGLE_GEMINI_2_5_FLASH_LITE_PREVIEW_09_2025,
@@ -172,6 +175,7 @@ var AITuberOnAirChat = (() => {
     MODEL_GPT_5_6_TERRA: () => MODEL_GPT_5_6_TERRA,
     MODEL_GPT_5_MINI: () => MODEL_GPT_5_MINI,
     MODEL_GPT_5_NANO: () => MODEL_GPT_5_NANO,
+    MODEL_GPT_6_1_SOL: () => MODEL_GPT_6_1_SOL,
     MODEL_GPT_6_ASTRA: () => MODEL_GPT_6_ASTRA,
     MODEL_GPT_6_LUNA: () => MODEL_GPT_6_LUNA,
     MODEL_GPT_6_SOL: () => MODEL_GPT_6_SOL,
@@ -225,8 +229,11 @@ var AITuberOnAirChat = (() => {
     MODEL_OPENAI_GPT_5_6_TERRA: () => MODEL_OPENAI_GPT_5_6_TERRA,
     MODEL_OPENAI_GPT_5_MINI: () => MODEL_OPENAI_GPT_5_MINI,
     MODEL_OPENAI_GPT_5_NANO: () => MODEL_OPENAI_GPT_5_NANO,
+    MODEL_OPENAI_GPT_6_1_SOL: () => MODEL_OPENAI_GPT_6_1_SOL,
     MODEL_OPENAI_GPT_6_ASTRA: () => MODEL_OPENAI_GPT_6_ASTRA,
     MODEL_OPENAI_GPT_6_ASTRA_PRO: () => MODEL_OPENAI_GPT_6_ASTRA_PRO,
+    MODEL_OPENAI_GPT_6_LUNA: () => MODEL_OPENAI_GPT_6_LUNA,
+    MODEL_OPENAI_GPT_6_SOL: () => MODEL_OPENAI_GPT_6_SOL,
     MODEL_OPENAI_GPT_LATEST: () => MODEL_OPENAI_GPT_LATEST,
     MODEL_OPENAI_GPT_MINI_LATEST: () => MODEL_OPENAI_GPT_MINI_LATEST,
     MODEL_OPENROUTER_AUTO: () => MODEL_OPENROUTER_AUTO,
@@ -244,6 +251,7 @@ var AITuberOnAirChat = (() => {
     MODEL_SAKANA_NAMAZU: () => MODEL_SAKANA_NAMAZU,
     MODEL_XAI_GROK_4_5: () => MODEL_XAI_GROK_4_5,
     MODEL_XAI_GROK_4_6: () => MODEL_XAI_GROK_4_6,
+    MODEL_XAI_GROK_4_7: () => MODEL_XAI_GROK_4_7,
     MODEL_XAI_GROK_LATEST: () => MODEL_XAI_GROK_LATEST,
     MODEL_ZAI_GLM_4_5_AIR: () => MODEL_ZAI_GLM_4_5_AIR,
     MODEL_ZAI_GLM_4_5_AIR_FREE: () => MODEL_ZAI_GLM_4_5_AIR_FREE,
@@ -251,6 +259,7 @@ var AITuberOnAirChat = (() => {
     MODEL_ZAI_GLM_5_2: () => MODEL_ZAI_GLM_5_2,
     MODEL_ZAI_GLM_5_3: () => MODEL_ZAI_GLM_5_3,
     MODEL_ZAI_GLM_5_3_FLASH: () => MODEL_ZAI_GLM_5_3_FLASH,
+    MODEL_ZAI_GLM_5_3_FLASHX: () => MODEL_ZAI_GLM_5_3_FLASHX,
     MistralChatService: () => MistralChatService,
     MistralChatServiceProvider: () => MistralChatServiceProvider,
     OPENAI_COMPATIBLE_LOCAL_PRESETS: () => OPENAI_COMPATIBLE_LOCAL_PRESETS,
@@ -362,6 +371,7 @@ var AITuberOnAirChat = (() => {
   // src/constants/openai.ts
   var ENDPOINT_OPENAI_CHAT_COMPLETIONS_API = "https://api.openai.com/v1/chat/completions";
   var ENDPOINT_OPENAI_RESPONSES_API = "https://api.openai.com/v1/responses";
+  var MODEL_GPT_6_1_SOL = "gpt-6.1-sol";
   var MODEL_GPT_6_ASTRA = "gpt-6-astra";
   var MODEL_GPT_6_SOL = "gpt-6-sol";
   var MODEL_GPT_6_LUNA = "gpt-6-luna";
@@ -387,6 +397,7 @@ var AITuberOnAirChat = (() => {
   var MODEL_O1_MINI = "o1-mini";
   var MODEL_O1 = "o1";
   var VISION_SUPPORTED_MODELS = [
+    MODEL_GPT_6_1_SOL,
     MODEL_GPT_6_ASTRA,
     MODEL_GPT_6_SOL,
     MODEL_GPT_6_LUNA,
@@ -430,16 +441,16 @@ var AITuberOnAirChat = (() => {
     return GPT_5_MODELS.includes(model);
   }
   function isOpenAIReasoningModel(model) {
-    return model === MODEL_GPT_6_ASTRA || model === MODEL_GPT_6_SOL || model === MODEL_GPT_6_LUNA || isGPT5Model(model);
+    return model === MODEL_GPT_6_1_SOL || model === MODEL_GPT_6_ASTRA || model === MODEL_GPT_6_SOL || model === MODEL_GPT_6_LUNA || isGPT5Model(model);
   }
   function isResponsesOnlyGPT5Model(model) {
     return model === MODEL_GPT_5_4_PRO;
   }
   function allowsReasoningXHigh(model) {
-    return model === MODEL_GPT_6_ASTRA || model === MODEL_GPT_6_SOL || model === MODEL_GPT_6_LUNA || model === MODEL_GPT_5_6 || model === MODEL_GPT_5_6_SOL || model === MODEL_GPT_5_6_TERRA || model === MODEL_GPT_5_6_LUNA || model === MODEL_GPT_5_5 || model === MODEL_GPT_5_4 || model === MODEL_GPT_5_4_MINI || model === MODEL_GPT_5_4_NANO || model === MODEL_GPT_5_4_PRO;
+    return model === MODEL_GPT_6_1_SOL || model === MODEL_GPT_6_ASTRA || model === MODEL_GPT_6_SOL || model === MODEL_GPT_6_LUNA || model === MODEL_GPT_5_6 || model === MODEL_GPT_5_6_SOL || model === MODEL_GPT_5_6_TERRA || model === MODEL_GPT_5_6_LUNA || model === MODEL_GPT_5_5 || model === MODEL_GPT_5_4 || model === MODEL_GPT_5_4_MINI || model === MODEL_GPT_5_4_NANO || model === MODEL_GPT_5_4_PRO;
   }
   function allowsReasoningMax(model) {
-    return model === MODEL_GPT_6_ASTRA || model === MODEL_GPT_6_SOL || model === MODEL_GPT_6_LUNA || model === MODEL_GPT_5_6 || model === MODEL_GPT_5_6_SOL || model === MODEL_GPT_5_6_TERRA || model === MODEL_GPT_5_6_LUNA;
+    return model === MODEL_GPT_6_1_SOL || model === MODEL_GPT_6_ASTRA || model === MODEL_GPT_6_SOL || model === MODEL_GPT_6_LUNA || model === MODEL_GPT_5_6 || model === MODEL_GPT_5_6_SOL || model === MODEL_GPT_5_6_TERRA || model === MODEL_GPT_5_6_LUNA;
   }
   function allowsReasoningNone(model) {
     return model === MODEL_GPT_6_SOL || model === MODEL_GPT_6_LUNA || model === MODEL_GPT_5_6 || model === MODEL_GPT_5_6_SOL || model === MODEL_GPT_5_6_TERRA || model === MODEL_GPT_5_6_LUNA || model === MODEL_GPT_5_1 || model === MODEL_GPT_5_4 || model === MODEL_GPT_5_5 || model === MODEL_GPT_5_4_MINI || model === MODEL_GPT_5_4_NANO;
@@ -460,7 +471,7 @@ var AITuberOnAirChat = (() => {
     return "medium";
   }
   function getDefaultReasoningEffortForOpenAIModel(model) {
-    return model === MODEL_GPT_6_ASTRA ? "low" : model === MODEL_GPT_6_SOL ? "medium" : model === MODEL_GPT_6_LUNA ? "low" : getDefaultReasoningEffortForGPT5Model(model);
+    return model === MODEL_GPT_6_1_SOL || model === MODEL_GPT_6_ASTRA ? "low" : model === MODEL_GPT_6_SOL ? "medium" : model === MODEL_GPT_6_LUNA ? "low" : getDefaultReasoningEffortForGPT5Model(model);
   }
 
   // src/constants/gemini.ts
@@ -651,6 +662,12 @@ var AITuberOnAirChat = (() => {
 
   // src/constants/openrouter.ts
   var ENDPOINT_OPENROUTER_API = "https://openrouter.ai/api/v1/chat/completions";
+  var MODEL_OPENAI_GPT_6_1_SOL = "openai/gpt-6.1-sol";
+  var MODEL_OPENAI_GPT_6_SOL = "openai/gpt-6-sol";
+  var MODEL_OPENAI_GPT_6_LUNA = "openai/gpt-6-luna";
+  var MODEL_ANTHROPIC_CLAUDE_SONNET_5_5 = "anthropic/claude-sonnet-5.5";
+  var MODEL_ANTHROPIC_CLAUDE_OPUS_5_5 = "anthropic/claude-opus-5.5";
+  var MODEL_XAI_GROK_4_7 = "x-ai/grok-4.7";
   var MODEL_OPENAI_GPT_6_ASTRA = "openai/gpt-6-astra";
   var MODEL_OPENAI_GPT_6_ASTRA_PRO = "openai/gpt-6-astra-pro";
   var MODEL_ANTHROPIC_CLAUDE_FABLE_5_1 = "anthropic/claude-fable-5.1";
@@ -707,6 +724,7 @@ var AITuberOnAirChat = (() => {
   var MODEL_ZAI_GLM_5_2 = "z-ai/glm-5.2";
   var MODEL_ZAI_GLM_5_3 = "z-ai/glm-5.3";
   var MODEL_ZAI_GLM_5_3_FLASH = "z-ai/glm-5.3-flash";
+  var MODEL_ZAI_GLM_5_3_FLASHX = "z-ai/glm-5.3-flashx";
   var MODEL_ZAI_GLM_4_7_FLASH = "z-ai/glm-4.7-flash";
   var MODEL_ZAI_GLM_4_5_AIR = "z-ai/glm-4.5-air";
   var MODEL_ZAI_GLM_4_5_AIR_FREE = "z-ai/glm-4.5-air:free";
@@ -769,6 +787,31 @@ var AITuberOnAirChat = (() => {
     "max"
   ];
   var OPENROUTER_MODEL_REASONING = {
+    // Chat-oriented defaults avoid disabling mandatory reasoning and favor latency.
+    [MODEL_OPENAI_GPT_6_1_SOL]: {
+      efforts: ["low", "medium", "high", "xhigh", "max"],
+      defaultEffort: "low"
+    },
+    [MODEL_OPENAI_GPT_6_SOL]: {
+      efforts: ["none", "low", "medium", "high", "xhigh", "max"],
+      defaultEffort: "none"
+    },
+    [MODEL_OPENAI_GPT_6_LUNA]: {
+      efforts: ["none", "low", "medium", "high", "xhigh", "max"],
+      defaultEffort: "none"
+    },
+    [MODEL_ANTHROPIC_CLAUDE_SONNET_5_5]: {
+      efforts: ["low", "medium", "high", "xhigh", "max"],
+      defaultEffort: "low"
+    },
+    [MODEL_ANTHROPIC_CLAUDE_OPUS_5_5]: {
+      efforts: ["low", "medium", "high", "xhigh", "max"],
+      defaultEffort: "low"
+    },
+    [MODEL_XAI_GROK_4_7]: {
+      efforts: ["low", "medium", "high", "xhigh"],
+      defaultEffort: "low"
+    },
     [MODEL_OPENAI_GPT_6_ASTRA]: {
       efforts: ["max", "xhigh", "high", "medium", "low"],
       defaultEffort: "medium"
@@ -821,6 +864,12 @@ var AITuberOnAirChat = (() => {
     MODEL_NEX_AGI_NEX_N2_5_PRO_FREE
   ];
   var OPENROUTER_VISION_SUPPORTED_MODELS = [
+    MODEL_OPENAI_GPT_6_1_SOL,
+    MODEL_OPENAI_GPT_6_SOL,
+    MODEL_OPENAI_GPT_6_LUNA,
+    MODEL_ANTHROPIC_CLAUDE_SONNET_5_5,
+    MODEL_ANTHROPIC_CLAUDE_OPUS_5_5,
+    MODEL_XAI_GROK_4_7,
     MODEL_OPENAI_GPT_6_ASTRA,
     MODEL_OPENAI_GPT_6_ASTRA_PRO,
     MODEL_ANTHROPIC_CLAUDE_FABLE_5_1,
@@ -866,6 +915,7 @@ var AITuberOnAirChat = (() => {
     MODEL_GOOGLE_GEMINI_2_5_FLASH,
     MODEL_GOOGLE_GEMINI_2_5_FLASH_LITE_PREVIEW_09_2025,
     MODEL_ZAI_GLM_5_3_FLASH,
+    MODEL_ZAI_GLM_5_3_FLASHX,
     MODEL_OPENROUTER_DEEPSEEK_V4_FLASH_VISION_EXP,
     MODEL_QWEN_QWEN_3_8_FLASH,
     MODEL_MOONSHOTAI_KIMI_K2_7_CODE,
@@ -893,7 +943,7 @@ var AITuberOnAirChat = (() => {
       normalizedModel
     ) ? OPENROUTER_MODEL_REASONING[normalizedModel] : void 0;
     if (profile) return profile.efforts;
-    if (normalizedModel === MODEL_ZAI_GLM_5_3 || normalizedModel === MODEL_ZAI_GLM_5_3_FLASH) {
+    if (normalizedModel === MODEL_ZAI_GLM_5_3 || normalizedModel === MODEL_ZAI_GLM_5_3_FLASH || normalizedModel === MODEL_ZAI_GLM_5_3_FLASHX) {
       return OPENROUTER_ALWAYS_THINKING_REASONING_EFFORTS;
     }
     if (normalizedModel === MODEL_OPENROUTER_DEEPSEEK_V4_FLASH_VISION_EXP) {
@@ -920,7 +970,7 @@ var AITuberOnAirChat = (() => {
       normalizedModel
     ) ? OPENROUTER_MODEL_REASONING[normalizedModel] : void 0;
     if (profile) return profile.defaultEffort;
-    if (normalizedModel === MODEL_ZAI_GLM_5_3 || normalizedModel === MODEL_ZAI_GLM_5_3_FLASH) {
+    if (normalizedModel === MODEL_ZAI_GLM_5_3 || normalizedModel === MODEL_ZAI_GLM_5_3_FLASH || normalizedModel === MODEL_ZAI_GLM_5_3_FLASHX) {
       return "low";
     }
     return normalizedModel === MODEL_ZAI_GLM_5_2 || normalizedModel === MODEL_QWEN_QWEN_3_8_FLASH || normalizedModel === MODEL_MOONSHOTAI_KIMI_K2_6 || normalizedModel === MODEL_ANTHROPIC_CLAUDE_SONNET_5 || normalizedModel === MODEL_OPENROUTER_DEEPSEEK_V4_FLASH_VISION_EXP || normalizedModel === MODEL_OPENROUTER_DEEPSEEK_V4_FLASH || normalizedModel === MODEL_OPENROUTER_DEEPSEEK_V4_FLASH_0731 || normalizedModel === MODEL_OPENROUTER_DEEPSEEK_V4_PRO_0813 ? "none" : void 0;
@@ -939,7 +989,7 @@ var AITuberOnAirChat = (() => {
     if (!requested) {
       return void 0;
     }
-    if (normalizedModel === MODEL_ZAI_GLM_5_3 || normalizedModel === MODEL_ZAI_GLM_5_3_FLASH) {
+    if (normalizedModel === MODEL_ZAI_GLM_5_3 || normalizedModel === MODEL_ZAI_GLM_5_3_FLASH || normalizedModel === MODEL_ZAI_GLM_5_3_FLASHX) {
       if (requested === "high") {
         return "high";
       }
@@ -989,6 +1039,7 @@ var AITuberOnAirChat = (() => {
   var ENDPOINT_ZAI_CHAT_COMPLETIONS_API = "https://api.z.ai/api/paas/v4/chat/completions";
   var MODEL_GLM_5_3 = "glm-5.3";
   var MODEL_GLM_5_3_FLASH = "glm-5.3-flash";
+  var MODEL_GLM_5_3_FLASHX = "glm-5.3-flashx";
   var MODEL_GLM_5_2 = "glm-5.2";
   var MODEL_GLM_5_1 = "glm-5.1";
   var MODEL_GLM_5 = "glm-5";
@@ -1016,7 +1067,7 @@ var AITuberOnAirChat = (() => {
     "max"
   ];
   function isZaiAlwaysThinkingModel(model) {
-    return model === MODEL_GLM_5_3 || model === MODEL_GLM_5_3_FLASH;
+    return model === MODEL_GLM_5_3 || model === MODEL_GLM_5_3_FLASH || model === MODEL_GLM_5_3_FLASHX;
   }
   function getZaiSupportedReasoningEfforts(model) {
     if (isZaiAlwaysThinkingModel(model)) {
@@ -1056,6 +1107,7 @@ var AITuberOnAirChat = (() => {
     return "max";
   }
   var ZAI_VISION_SUPPORTED_MODELS = [
+    MODEL_GLM_5_3_FLASHX,
     MODEL_GLM_5_3_FLASH,
     MODEL_GLM_5V_TURBO,
     MODEL_GLM_4_6V,
@@ -2874,6 +2926,15 @@ If it's in another language, summarize in that language.
   };
 
   // src/services/providers/openai/responsesParser.ts
+  function throwIfResponsesError(data, eventType) {
+    const response = data?.response ?? data;
+    const error = response?.error;
+    if (!error && response?.status !== "failed" && eventType !== "error" && eventType !== "response.failed") {
+      return;
+    }
+    const message = typeof error === "string" ? error : error?.message ?? data?.message ?? "The provider response failed.";
+    throw new Error(`Provider response error: ${message}`);
+  }
   async function parseOpenAIResponsesStream(res, onPartial, options = {}) {
     const reader = res.body.getReader();
     const dec = new TextDecoder();
@@ -2884,12 +2945,12 @@ If it's in another language, summarize in that language.
     let incompleteDetails;
     let usage;
     let buf = "";
+    let eventType = "";
+    let eventData = "";
     while (true) {
       const { done, value } = await reader.read();
       if (done) break;
       buf += dec.decode(value, { stream: true });
-      let eventType = "";
-      let eventData = "";
       const lines = buf.split("\n");
       buf = lines.pop() || "";
       for (let i = 0; i < lines.length; i++) {
@@ -2899,31 +2960,36 @@ If it's in another language, summarize in that language.
         } else if (line.startsWith("data:")) {
           eventData = line.slice(5).trim();
         } else if (line === "" && eventType && eventData) {
+          let json;
           try {
-            const json = JSON.parse(eventData);
-            handleResponsesSSEEvent(
-              eventType,
-              json,
-              onPartial,
-              textBlocks,
-              toolCallsMap,
-              providerItems,
-              (metadata) => {
-                if (metadata.responseStatus !== void 0) {
-                  responseStatus = metadata.responseStatus;
-                }
-                if (metadata.incompleteDetails !== void 0) {
-                  incompleteDetails = metadata.incompleteDetails;
-                }
-                if (metadata.usage !== void 0) {
-                  usage = metadata.usage;
-                }
-              }
-            );
+            json = JSON.parse(eventData);
           } catch (error) {
             options.onJsonError?.(eventData, error);
             console.warn("Failed to parse SSE data:", eventData);
+            eventType = "";
+            eventData = "";
+            continue;
           }
+          throwIfResponsesError(json, eventType);
+          handleResponsesSSEEvent(
+            eventType,
+            json,
+            onPartial,
+            textBlocks,
+            toolCallsMap,
+            providerItems,
+            (metadata) => {
+              if (metadata.responseStatus !== void 0) {
+                responseStatus = metadata.responseStatus;
+              }
+              if (metadata.incompleteDetails !== void 0) {
+                incompleteDetails = metadata.incompleteDetails;
+              }
+              if (metadata.usage !== void 0) {
+                usage = metadata.usage;
+              }
+            }
+          );
           eventType = "";
           eventData = "";
         }
@@ -3051,6 +3117,7 @@ If it's in another language, summarize in that language.
     };
   }
   function parseOpenAIResponsesOneShot(data, options = {}) {
+    throwIfResponsesError(data);
     const blocks = [];
     if (data.output && Array.isArray(data.output)) {
       data.output.forEach((outputItem) => {
@@ -3185,10 +3252,15 @@ If it's in another language, summarize in that language.
     chatTemplateThinking,
     maxTokens
   }) {
-    if (model === MODEL_GPT_6_ASTRA && (!reasoning_effort || reasoning_effort === "none" || reasoning_effort === "minimal")) {
+    if ((model === MODEL_GPT_6_ASTRA || provider === "openai" && model === MODEL_GPT_6_1_SOL) && (!reasoning_effort || reasoning_effort === "none" || reasoning_effort === "minimal")) {
       reasoning_effort = "low";
     }
     const isResponsesAPI = endpoint === ENDPOINT_OPENAI_RESPONSES_API;
+    if (provider === "openai" && model === MODEL_GPT_6_1_SOL && (tools.length > 0 || mcpServers.length > 0) && !isResponsesAPI) {
+      throw new Error(
+        "GPT-6.1 Sol tool calling requires the OpenAI Responses API endpoint: " + ENDPOINT_OPENAI_RESPONSES_API + ". Custom endpoints use Chat Completions and cannot be used with this model's tools."
+      );
+    }
     validateMCPCompatibility(endpoint, mcpServers);
     const body = {
       model,
@@ -3405,7 +3477,8 @@ If it's in another language, summarize in that language.
       this.apiKey = apiKey;
       this.model = model;
       this.tools = tools || [];
-      this.endpoint = (model === MODEL_GPT_6_ASTRA || visionModel === MODEL_GPT_6_ASTRA) && endpoint === ENDPOINT_OPENAI_CHAT_COMPLETIONS_API ? ENDPOINT_OPENAI_RESPONSES_API : endpoint;
+      const solToolsRequireResponses = provider === "openai" && (model === MODEL_GPT_6_1_SOL || visionModel === MODEL_GPT_6_1_SOL) && (this.tools.length > 0 || mcpServers.length > 0);
+      this.endpoint = (model === MODEL_GPT_6_ASTRA || visionModel === MODEL_GPT_6_ASTRA || solToolsRequireResponses) && endpoint === ENDPOINT_OPENAI_CHAT_COMPLETIONS_API ? ENDPOINT_OPENAI_RESPONSES_API : endpoint;
       this.mcpServers = mcpServers;
       this.responseLength = responseLength;
       this.verbosity = verbosity;
@@ -3572,15 +3645,18 @@ If it's in another language, summarize in that language.
       return res;
     }
     async handleStream(res, onPartial) {
-      return parseOpenAICompatibleTextStream(res, onPartial);
+      return parseOpenAICompatibleTextStream(res, onPartial, {
+        throwOnApiError: true
+      });
     }
     async parseStream(res, onPartial) {
       return parseOpenAICompatibleToolStream(res, onPartial, {
-        appendTextBlock: StreamTextAccumulator.addTextBlock
+        appendTextBlock: StreamTextAccumulator.addTextBlock,
+        throwOnApiError: true
       });
     }
     parseOneShot(data) {
-      return parseOpenAICompatibleOneShot(data);
+      return parseOpenAICompatibleOneShot(data, { throwOnApiError: true });
     }
   };
 
@@ -5540,6 +5616,9 @@ If it's in another language, summarize in that language.
         shouldUseResponsesAPI = preference !== "chat" || optimizedOptions.reasoning_effort !== "none";
       } else if (isResponsesOnlyGPT5Model(modelName)) {
         shouldUseResponsesAPI = true;
+      } else if (modelName === MODEL_GPT_6_1_SOL || visionModel === MODEL_GPT_6_1_SOL) {
+        const preference = optimizedOptions.gpt5EndpointPreference || "responses";
+        shouldUseResponsesAPI = (tools?.length ?? 0) > 0 || preference !== "chat";
       } else if (isOpenAIReasoningModel(modelName)) {
         const preference = optimizedOptions.gpt5EndpointPreference || "chat";
         shouldUseResponsesAPI = preference === "responses";
@@ -5583,6 +5662,7 @@ If it's in another language, summarize in that language.
         MODEL_GPT_5_4,
         MODEL_GPT_5_5,
         MODEL_GPT_5_6,
+        MODEL_GPT_6_1_SOL,
         MODEL_GPT_6_ASTRA,
         MODEL_GPT_6_SOL,
         MODEL_GPT_6_LUNA,
@@ -5947,7 +6027,7 @@ If it's in another language, summarize in that language.
     async handleStream(res, onPartial, model) {
       const text = await parseOpenAICompatibleTextStream(res, onPartial, {
         onJsonError: (payload) => console.debug("Failed to parse SSE data:", payload),
-        throwOnApiError: isOpenRouterAutoModel(model)
+        throwOnApiError: true
       });
       ensureAutoRouterOutput(model, text.trim().length > 0);
       return text;
@@ -5958,7 +6038,7 @@ If it's in another language, summarize in that language.
     async parseStream(res, onPartial, model) {
       const completion = await parseOpenAICompatibleToolStream(res, onPartial, {
         onJsonError: (payload) => console.debug("Failed to parse SSE data:", payload),
-        throwOnApiError: isOpenRouterAutoModel(model)
+        throwOnApiError: true
       });
       ensureAutoRouterOutput(model, hasUsableCompletionOutput(completion));
       return completion;
@@ -5968,7 +6048,7 @@ If it's in another language, summarize in that language.
      */
     parseOneShot(data, model) {
       const completion = parseOpenAICompatibleOneShot(data, {
-        throwOnApiError: isOpenRouterAutoModel(model)
+        throwOnApiError: true
       });
       ensureAutoRouterOutput(model, hasUsableCompletionOutput(completion));
       return completion;
@@ -6022,6 +6102,12 @@ If it's in another language, summarize in that language.
      */
     getSupportedModels() {
       return [
+        MODEL_OPENAI_GPT_6_1_SOL,
+        MODEL_OPENAI_GPT_6_SOL,
+        MODEL_OPENAI_GPT_6_LUNA,
+        MODEL_ANTHROPIC_CLAUDE_SONNET_5_5,
+        MODEL_ANTHROPIC_CLAUDE_OPUS_5_5,
+        MODEL_XAI_GROK_4_7,
         MODEL_OPENAI_GPT_6_ASTRA,
         MODEL_OPENAI_GPT_6_ASTRA_PRO,
         MODEL_ANTHROPIC_CLAUDE_FABLE_5_1,
@@ -6075,6 +6161,7 @@ If it's in another language, summarize in that language.
         // Z.ai models
         MODEL_ZAI_GLM_5_3,
         MODEL_ZAI_GLM_5_3_FLASH,
+        MODEL_ZAI_GLM_5_3_FLASHX,
         MODEL_ZAI_GLM_5_2,
         MODEL_ZAI_GLM_4_7_FLASH,
         MODEL_ZAI_GLM_4_5_AIR,
@@ -6673,10 +6760,19 @@ If it's in another language, summarize in that language.
       if (this.responseFormat) {
         body.response_format = this.responseFormat;
       }
-      if (this.thinking) {
+      if (model === MODEL_GLM_5_3_FLASHX) {
+        body.thinking = {
+          type: "enabled",
+          clear_thinking: this.thinking?.clear_thinking ?? true
+        };
+        body.reasoning_effort = normalizeZaiReasoningEffort(
+          model,
+          this.reasoningEffort
+        );
+      } else if (this.thinking) {
         body.thinking = this.thinking;
       }
-      if (isZaiReasoningEffortModel(model) && this.reasoningEffort) {
+      if (model !== MODEL_GLM_5_3_FLASHX && isZaiReasoningEffortModel(model) && this.reasoningEffort) {
         body.reasoning_effort = this.reasoningEffort;
       }
       const tools = this.buildToolsDefinition();
@@ -6694,6 +6790,7 @@ If it's in another language, summarize in that language.
     }
     async handleStream(res, onPartial) {
       return parseOpenAICompatibleTextStream(res, onPartial, {
+        throwOnApiError: true,
         onJsonError: (payload) => console.debug("Failed to parse SSE data:", payload)
       });
     }
@@ -6702,6 +6799,7 @@ If it's in another language, summarize in that language.
      */
     async parseStream(res, onPartial) {
       return parseOpenAICompatibleToolStream(res, onPartial, {
+        throwOnApiError: true,
         onJsonError: (payload) => console.debug("Failed to parse SSE data:", payload)
       });
     }
@@ -6709,7 +6807,7 @@ If it's in another language, summarize in that language.
      * Parse non-streaming response
      */
     parseOneShot(data) {
-      return parseOpenAICompatibleOneShot(data);
+      return parseOpenAICompatibleOneShot(data, { throwOnApiError: true });
     }
   };
 
@@ -6778,6 +6876,7 @@ If it's in another language, summarize in that language.
       return [
         MODEL_GLM_5_3,
         MODEL_GLM_5_3_FLASH,
+        MODEL_GLM_5_3_FLASHX,
         MODEL_GLM_5_2,
         MODEL_GLM_5_1,
         MODEL_GLM_5,
@@ -7034,7 +7133,7 @@ If it's in another language, summarize in that language.
         mcp: MCP_SUPPORTED_PROVIDERS.has(providerName),
         jsonMode: JSON_MODE_SUPPORTED_PROVIDERS.has(providerName),
         responseLength: true,
-        reasoningEffort: providerName === "openai" && model === MODEL_GPT_6_ASTRA ? ["low", "medium", "high", "xhigh", "max"] : providerName === "openai" && model === MODEL_GPT_6_SOL ? ["none", "low", "medium", "high", "xhigh", "max"] : providerName === "openai" && model === MODEL_GPT_6_LUNA ? ["none", "low", "medium", "high", "xhigh", "max"] : providerName === "claude" && model ? [...getClaudeSupportedReasoningEfforts(model)] : providerName === "gemini" && model ? [...getGeminiSupportedReasoningEfforts(model)] : providerName === "kimi" && model ? [...getKimiSupportedReasoningEfforts(model)] : providerName === "deepseek" && model ? [...getDeepSeekSupportedReasoningEfforts(model)] : providerName === "openrouter" && model ? [...getOpenRouterSupportedReasoningEfforts(model)] : providerName === "xai" && model ? [...getXaiSupportedReasoningEfforts(model)] : providerName === "zai" && model ? [...getZaiSupportedReasoningEfforts(model)] : REASONING_EFFORT_BY_PROVIDER[providerName] ?? []
+        reasoningEffort: providerName === "openai" && (model === MODEL_GPT_6_1_SOL || model === MODEL_GPT_6_ASTRA) ? ["low", "medium", "high", "xhigh", "max"] : providerName === "openai" && model === MODEL_GPT_6_SOL ? ["none", "low", "medium", "high", "xhigh", "max"] : providerName === "openai" && model === MODEL_GPT_6_LUNA ? ["none", "low", "medium", "high", "xhigh", "max"] : providerName === "claude" && model ? [...getClaudeSupportedReasoningEfforts(model)] : providerName === "gemini" && model ? [...getGeminiSupportedReasoningEfforts(model)] : providerName === "kimi" && model ? [...getKimiSupportedReasoningEfforts(model)] : providerName === "deepseek" && model ? [...getDeepSeekSupportedReasoningEfforts(model)] : providerName === "openrouter" && model ? [...getOpenRouterSupportedReasoningEfforts(model)] : providerName === "xai" && model ? [...getXaiSupportedReasoningEfforts(model)] : providerName === "zai" && model ? [...getZaiSupportedReasoningEfforts(model)] : REASONING_EFFORT_BY_PROVIDER[providerName] ?? []
       };
     }
     /**