@bman654/clodex 2.18.3 → 2.18.5

package/dist/cli.js

@@ -346,7 +346,7 @@ import { join } from "path";
 // package.json
 var package_default = {
   name: "@bman654/clodex",
-  version: "2.18.3",
+  version: "2.18.5",
   publishConfig: {
     access: "public"
   },
@@ -433,7 +433,7 @@ var package_default = {
 
 // src/constants.ts
 var CODEX_RESPONSES_LITE_WS_URL = "wss://chatgpt.com/backend-api/codex/responses";
-var CODEX_RESPONSES_LITE_VERSION = "0.156.0";
+var CODEX_RESPONSES_LITE_VERSION = "0.159.0";
 var CODEX_RESPONSES_WEBSOCKETS_BETA = "responses_websockets=2026-02-06";
 var TEST_TIMEOUT_MS = 1e4;
 var CONFLICTING_ENV_VARS = [
@@ -630,6 +630,7 @@ var HEURISTIC_RULES = [
   [/claude/i, 2e5],
   [/deepseek-v4|deepseek-r1|deepseek-reasoner/i, 1e6],
   [/deepseek/i, 64e3],
+  [/gpt-6\.1-sol/i, 105e4],
   [/gpt-6-(?:astra|sol|luna)/i, 105e4],
   [/gpt-5|gpt-4\.1|o3-|o4-/i, 1e6],
   [/gpt-4o|gpt-4-turbo|gpt-4/i, 128e3],
@@ -4916,31 +4917,6 @@ var opencode_go_models_default = [
     upstreamModelId: "deepseek-v4.1-flash",
     family: "deepseek"
   },
-  {
-    id: "glm-5.1",
-    name: "GLM-5.1",
-    contextWindow: 202752,
-    cost: {
-      input: 1.4,
-      output: 4.4,
-      cache_read: 0.26
-    },
-    modelFormat: "openai",
-    npm: "@ai-sdk/openai-compatible",
-    apiUrl: "https://opencode.ai/zen/go/v1",
-    reasoning: true,
-    modalities: [
-      "text"
-    ],
-    compatibility: {
-      supportsReasoningEffort: false,
-      supportsStore: false,
-      supportsDeveloperRole: false,
-      maxTokensField: "max_tokens"
-    },
-    upstreamModelId: "glm-5.1",
-    family: "glm"
-  },
   {
     id: "glm-5.2",
     name: "GLM-5.2",
@@ -5009,6 +4985,39 @@ var opencode_go_models_default = [
     upstreamModelId: "gpt-5.6-luna",
     family: "gpt"
   },
+  {
+    id: "gpt-6-luna",
+    name: "GPT-6 Luna",
+    contextWindow: 105e4,
+    pricingBoundary: 272e3,
+    pricingBoundaryNote: "Above it, OpenCode Go lists $0.20 input and $0.75 output per million tokens.",
+    cost: {
+      input: 0.1,
+      output: 0.5,
+      cache_read: 0.01,
+      cache_write: 0.125
+    },
+    modelFormat: "openai",
+    npm: "@ai-sdk/openai",
+    apiUrl: "https://opencode.ai/zen/go/v1",
+    reasoning: true,
+    modalities: [
+      "text",
+      "image"
+    ],
+    compatibility: {
+      reasoningEffortMap: {
+        none: "none",
+        low: "low",
+        medium: "medium",
+        high: "high",
+        xhigh: "xhigh",
+        max: "max"
+      }
+    },
+    upstreamModelId: "gpt-6-luna",
+    family: "gpt"
+  },
   {
     id: "hy3",
     name: "Hy3",
@@ -5193,7 +5202,8 @@ var opencode_go_models_default = [
     cost: {
       input: 0.3,
       output: 1.2,
-      cache_read: 0.06
+      cache_read: 0.06,
+      cache_write: 0.375
     },
     modelFormat: "openai",
     npm: "@ai-sdk/openai-compatible",
@@ -6072,6 +6082,13 @@ var CHATGPT_CODEX_UNSUPPORTED_MODELS = /* @__PURE__ */ new Set([
   // confirmed: rejected by chatgpt.com/backend-api/codex
 ]);
 var OPENAI_OAUTH_MODEL_SEEDS = [
+  // GPT-6.1 Sol: the 272K window and 872K ceiling are what the live Codex catalog
+  // returned on 2026-09-29; a request near 872K has not been tried. The output limit
+  // is the model card's. The catalog reports minimal_client_version 0.153.0, but on a
+  // Plus account versions through 0.158.0 were refused and 0.159.0 worked (#298), so
+  // this seed records that observed gate. A live refresh stores the catalog's value
+  // instead. CODEX_RESPONSES_LITE_VERSION must stay at or above this.
+  { id: "gpt-6.1-sol", name: "GPT-6.1 Sol", contextWindow: 272e3, maxContextWindow: 872e3, maxOutputTokens: 128e3, reasoning: true, useResponsesLite: true, preferWebSockets: true, minimalClientVersion: "0.159.0" },
   // GPT-6 family. The window, ceiling and Responses-Lite flags are what the live
   // Codex catalog returned (Astra on 2026-09-04, Sol and Luna on 2026-09-22) and are
   // deliberately NOT the published API numbers: the model cards list a 1,050,000
@@ -13025,8 +13042,8 @@ function translateRequest(body, npm, options) {
   }
   const upstreamModelId2 = options?.reasoningMetadata?.upstreamModelId ?? body.model;
   const supportsExplicitOpenAiCaching = !options?.openAiOAuth && supportsOpenAiPromptCacheBreakpoints(upstreamModelId2);
+  const claudeSessionId = extractClaudeSessionId(body, options?.claudeSessionId);
   if (npm === "@ai-sdk/openai") {
-    const claudeSessionId = extractClaudeSessionId(body, options?.claudeSessionId);
     const serviceTier = options?.openAiOAuth ? oauthServiceTier() : void 0;
     providerOptions = deepMergeProviderOptions(providerOptions, {
       openai: {
@@ -13047,7 +13064,17 @@ function translateRequest(body, npm, options) {
     toolChoice: compactRequest ? "none" : translateToolChoice(body.tool_choice),
     maxOutputTokens: options?.openAiOAuth ? void 0 : body.max_tokens,
     temperature: body.temperature,
-    providerOptions
+    providerOptions,
+    // OpenRouter derives its own conversation key when none is sent, and routes a
+    // session's requests to one provider when one is. The value needs to be stable
+    // and unique, not recognizable, so the session UUID goes over hashed through
+    // the same key the OpenAI route's prompt_cache_key uses; the system/tools hash
+    // stays the fallback for clients that send no session identity.
+    ...isOpenRouterRoute(npm, options?.reasoningMetadata) ? {
+      headers: {
+        "x-session-id": claudeSessionId ? claudeSessionPromptCacheKey(claudeSessionId) : openAiPromptCacheKey(baseSystem, upstreamTools)
+      }
+    } : {}
   };
 }
 function isOpenAiOAuthRoute(route) {