aieo 0.1.39 → 0.1.41

package/dist/index.d.cts

@@ -125,7 +125,7 @@ declare function gatewayEnvForProviders(baseUrl: string): {
  */
 declare const normalizeCallerBaseURL: typeof gatewayUrlFor;
 declare const PROVIDERS: Provider[];
-type ModelName = "sonnet" | "opus" | "haiku" | "gemini" | "gpt" | "kimi" | "grok";
+type ModelName = "sonnet" | "opus" | "haiku" | "gemini" | "gpt" | "kimi" | "glm" | "grok";
 type ModelId = string;
 declare const MODELS: Record<Provider, Partial<Record<ModelName, ModelId>>>;
 declare const DEFAULT_MODELS: Record<Provider, string>;

package/dist/index.d.ts

@@ -125,7 +125,7 @@ declare function gatewayEnvForProviders(baseUrl: string): {
  */
 declare const normalizeCallerBaseURL: typeof gatewayUrlFor;
 declare const PROVIDERS: Provider[];
-type ModelName = "sonnet" | "opus" | "haiku" | "gemini" | "gpt" | "kimi" | "grok";
+type ModelName = "sonnet" | "opus" | "haiku" | "gemini" | "gpt" | "kimi" | "glm" | "grok";
 type ModelId = string;
 declare const MODELS: Record<Provider, Partial<Record<ModelName, ModelId>>>;
 declare const DEFAULT_MODELS: Record<Provider, string>;

package/dist/index.js

@@ -23087,20 +23087,23 @@ var PROVIDERS = [
 var MODELS = {
   anthropic: {
     sonnet: "claude-sonnet-5",
-    opus: "claude-opus-4-6",
+    opus: "claude-opus-5",
     haiku: "claude-haiku-4-5"
   },
   google: {
-    gemini: "gemini-3-pro-preview"
+    gemini: "gemini-3.8-flash"
   },
   openai: {
-    gpt: "gpt-5"
+    gpt: "gpt-5.6-luna"
   },
   openrouter: {
-    kimi: "moonshotai/kimi-k2.6"
+    kimi: "moonshotai/kimi-k3",
+    // OpenRouter rolling alias (the leading "~" is part of the id): always
+    // the newest GLM Flash. The plain "z-ai/glm-flash-latest" 404s.
+    glm: "~z-ai/glm-flash-latest"
   },
   xai: {
-    grok: "grok-4"
+    grok: "grok-4.6"
   }
 };
 var DEFAULT_MODELS = {
@@ -23139,6 +23142,7 @@ function getProviderForModel(modelName) {
   }
   switch (modelName) {
     case "kimi":
+    case "glm":
       return "openrouter";
     case "sonnet":
       return "anthropic";
@@ -23154,13 +23158,19 @@ function getProviderForModel(modelName) {
       return "xai";
     // Full model IDs
     case "claude-sonnet-5":
+    case "claude-opus-5":
     case "claude-opus-4-6":
     case "claude-haiku-4-5":
       return "anthropic";
+    case "gemini-3.8-flash":
     case "gemini-3-pro-preview":
     case "gemini-2.0-flash":
       return "google";
     case "gpt-5":
+    case "gpt-5.5":
+    case "gpt-5.6-luna":
+    case "gpt-5.6-sol":
+    case "gpt-5.6-terra":
     case "gpt-4.1-mini":
       return "openai";
     default:
@@ -23170,6 +23180,11 @@ function getProviderForModel(modelName) {
       if (process.env.LLM_PROVIDER && PROVIDERS.includes(process.env.LLM_PROVIDER)) {
         return process.env.LLM_PROVIDER;
       }
+      if (typeof modelName === "string") {
+        const m = modelName.toLowerCase();
+        if (m.startsWith("gpt-")) return "openai";
+        if (m.startsWith("gemini-")) return "google";
+      }
       return "anthropic";
   }
 }
@@ -23338,6 +23353,7 @@ function getModel(provider, opts) {
       "gemini",
       "gpt",
       "kimi",
+      "glm",
       "grok"
     ];
     if (knownShortcuts.includes(opts.modelName)) {
@@ -23463,29 +23479,37 @@ function getModel(provider, opts) {
   }
 }
 var MODEL_CONTEXT_LIMITS = {
-  // Anthropic
+  // Anthropic — 1M is the default and the max on the 5-series; Haiku 4.5 is 200k.
   "claude-sonnet-5": 1e6,
+  "claude-opus-5": 1e6,
   "claude-opus-4-6": 1e6,
   "claude-haiku-4-5": 2e5,
-  // Google
-  "gemini-3-pro-preview": 1e6,
-  "gemini-2.0-flash": 1e6,
-  // OpenAI
+  // Google — Gemini's "1M" is 2^20.
+  "gemini-3.8-flash": 1048576,
+  "gemini-3-pro-preview": 1048576,
+  "gemini-2.0-flash": 1048576,
+  // OpenAI — 5.4+ is 1,050,000; the 4.1 family is 1,047,576; 5 through 5.3 are 400k.
   "gpt-5": 4e5,
   "gpt-5.5": 105e4,
   "gpt-5.6-terra": 105e4,
   "gpt-5.6-luna": 105e4,
   "gpt-5.6-sol": 105e4,
-  "gpt-4.1-mini": 1e6,
+  "gpt-4.1-mini": 1047576,
   // OpenRouter — values from the OpenRouter model catalog
   // (https://openrouter.ai/api/v1/models, context_length).
   "stealth/ox-alpha": 1048576,
+  "anthropic/claude-opus-5": 1e6,
+  "anthropic/claude-sonnet-5": 1e6,
   "openai/gpt-5": 4e5,
   "openai/gpt-5.5": 105e4,
   "openai/gpt-5.6-terra": 105e4,
   "openai/gpt-5.6-luna": 105e4,
   "openai/gpt-5.6-sol": 105e4,
   "moonshotai/kimi-k3": 1048576,
+  // Z.ai via OpenRouter. "~…-latest" is OpenRouter's rolling-alias form.
+  "~z-ai/glm-flash-latest": 1310720,
+  "z-ai/glm-5.3-flash": 1310720,
+  "z-ai/glm-5.3": 1310720,
   "moonshotai/kimi-k2.7-code": 262144,
   "moonshotai/kimi-k2.6": 262144,
   "moonshotai/kimi-k2.5": 262144,
@@ -23495,11 +23519,17 @@ var MODEL_CONTEXT_LIMITS = {
   // right matters beyond display: contextLimit drives truncateOldToolResults,
   // and premature truncation rewrites old messages — which invalidates Grok's
   // automatic prefix cache on every subsequent step.
+  "grok-4.6": 5e5,
+  "grok-4.5": 5e5,
+  "grok-4.3": 1e6,
   "grok-4": 256e3,
   "grok-4-fast": 2e6,
   "grok-4-fast-reasoning": 2e6,
   "grok-4-fast-non-reasoning": 2e6,
   "grok-code-fast-1": 256e3,
+  "x-ai/grok-4.6": 5e5,
+  "x-ai/grok-4.5": 5e5,
+  "x-ai/grok-4.3": 1e6,
   "x-ai/grok-4": 256e3,
   "x-ai/grok-4-fast": 2e6,
   "x-ai/grok-code-fast-1": 256e3