nolo-cli 0.70.0-alpha.1 → 0.70.0

package/{chunk-AJHEFRGJ.js → chunk-5YIL5WDC.js}

@@ -32,12 +32,7 @@ var resolveFireworksKimiModel = (model) => {
 var PLATFORM_HOSTED_CLAUDE_SONNET_5_MODEL = "anthropic/claude-sonnet-5";
 var PLATFORM_HOSTED_CLAUDE_OPUS_5_MODEL = "anthropic/claude-opus-5";
 var PLATFORM_HOSTED_CLAUDE_FABLE_5_MODEL = "anthropic/claude-fable-5";
-var PLATFORM_HOSTED_CLAUDE_HAIKU_45_MODEL = "anthropic/claude-haiku-4-5";
 var PLATFORM_HOSTED_GROK_4_6_MODEL = "grok-4.6";
-var PLATFORM_HOSTED_GPT_6_ASTRA_MODEL = "gpt-6-astra";
-var PLATFORM_HOSTED_GPT_56_SOL_MODEL = "gpt-5.6-sol";
-var PLATFORM_HOSTED_GPT_56_TERRA_MODEL = "gpt-5.6-terra";
-var PLATFORM_HOSTED_GPT_56_LUNA_MODEL = "gpt-5.6-luna";
 var PLATFORM_HOSTED_KIMI_K26_OPENROUTER_MODEL_ID = "qwen/qwen3.8-27b";
 var PLATFORM_HOSTED_GLM_53_MODEL = "glm-5.3";
 var PLATFORM_HOSTED_GLM_52_MODEL = "glm-5.2";
@@ -54,6 +49,7 @@ var PLATFORM_HOSTED_DEEPSEEK_FLASH_MODEL = "deepseek-flash";
 var PLATFORM_HOSTED_DEEPSEEK_FLASH_VISION_EXP_MODEL = "deepseek-v4-flash-vision-exp";
 var PLATFORM_HOSTED_DEEPSEEK_PRO_MODEL = "deepseek-v4-pro";
 var PLATFORM_HOSTED_LEGACY_DEEPSEEK_V4_FLASH_MODEL = "deepseek-v4-flash";
+var PLATFORM_HOSTED_NEMOTRON_35_LIGHTNING_MODEL = "nemotron-3-5-lightning-30b";
 var PLATFORM_HOSTED_KIMI_K3_MIN_CLIENT_VERSION = "0.38.0-alpha.4";
 var PLATFORM_HOSTED_GLM_53_FLASH_MIN_CLIENT_VERSION = "0.33.0-alpha.1";
 var PLATFORM_HOSTED_ROUTING_TABLE = {
@@ -65,22 +61,18 @@ var PLATFORM_HOSTED_ROUTING_TABLE = {
     wire: "chat.completions",
     agentRunHosted: true
   },
-  // GLM 5.3 & GLM 5.2 -> Baseten（inference.baseten.co）
   [PLATFORM_HOSTED_GLM_53_MODEL]: {
-    endpoint: "https://inference.baseten.co/v1/chat/completions",
-    usageProvider: "baseten",
-    keyName: "baseten",
-    // Baseten model library 的官方 id
-    upstreamModelId: "zai-org/GLM-5.3",
+    endpoint: "https://crof.ai/v1/chat/completions",
+    usageProvider: "upstream-k3",
+    keyName: "upstream-k3",
     wire: "chat.completions",
     agentRunHosted: true
   },
   [PLATFORM_HOSTED_GLM_52_MODEL]: {
-    endpoint: "https://inference.baseten.co/v1/chat/completions",
-    usageProvider: "baseten",
-    keyName: "baseten",
-    // legacy 5.2 显式 remap 到 baseten 的 zai-org/GLM-5.3，不得透传 "glm-5.2"
-    upstreamModelId: "zai-org/GLM-5.3",
+    endpoint: "https://crof.ai/v1/chat/completions",
+    usageProvider: "upstream-k3",
+    keyName: "upstream-k3",
+    upstreamModelId: PLATFORM_HOSTED_GLM_53_MODEL,
     wire: "chat.completions",
     agentRunHosted: true
   },
@@ -101,13 +93,11 @@ var PLATFORM_HOSTED_ROUTING_TABLE = {
     agentRunHosted: true,
     minClientVersion: PLATFORM_HOSTED_GLM_53_FLASH_MIN_CLIENT_VERSION
   },
-  // Kimi K3 -> Baseten（inference.baseten.co；model library 的 moonshotai/Kimi-K3）
   // wire 要求专属 body quirk（删采样参数 + max_completion_tokens），旧客户端本地直连必断。
   [PLATFORM_HOSTED_KIMI_K3_MODEL]: {
-    endpoint: "https://inference.baseten.co/v1/chat/completions",
-    usageProvider: "baseten",
-    keyName: "baseten",
-    upstreamModelId: "moonshotai/Kimi-K3",
+    endpoint: "https://crof.ai/v1/chat/completions",
+    usageProvider: "upstream-k3",
+    keyName: "upstream-k3",
     wire: "chat.completions",
     agentRunHosted: true,
     minClientVersion: PLATFORM_HOSTED_KIMI_K3_MIN_CLIENT_VERSION
@@ -166,70 +156,32 @@ var PLATFORM_HOSTED_ROUTING_TABLE = {
     wire: "chat.completions",
     agentRunHosted: true
   },
-  // Claude 系（真实模型）：DeepInfra 官方 id 与平台 id 同名，无需重映射。
-  // DeepInfra 这批模型无缓存价（rate_per_input_token_cached=null），计费侧对
-  // 无 inputCacheHit 的 deepinfra 模型按 input 全价计（见 calculatePrice）。
+  // Claude 系已下架（2026-09-01）：旧模型名保留路由做兼容，一律重映射到
+  // RunInfra 的 glm-5-3-flash（同 kimi-k2.6 → qwen 先例）。存量 agent 记录由
+  // modelUpgradeTable 迁移，这里兜住迁移窗口期与漏网请求。
   [PLATFORM_HOSTED_CLAUDE_SONNET_5_MODEL]: {
-    endpoint: "https://api.deepinfra.com/v1/openai/chat/completions",
-    usageProvider: "deepinfra",
-    keyName: "deepinfra",
-    upstreamModelId: PLATFORM_HOSTED_CLAUDE_SONNET_5_MODEL,
+    endpoint: "https://api.runinfra.ai/v1/chat/completions",
+    usageProvider: "runinfra",
+    keyName: "runinfra",
     wire: "chat.completions",
-    agentRunHosted: true
+    agentRunHosted: true,
+    upstreamModelId: PLATFORM_HOSTED_GLM_53_FLASH_MODEL
   },
   [PLATFORM_HOSTED_CLAUDE_OPUS_5_MODEL]: {
-    endpoint: "https://api.deepinfra.com/v1/openai/chat/completions",
-    usageProvider: "deepinfra",
-    keyName: "deepinfra",
-    upstreamModelId: PLATFORM_HOSTED_CLAUDE_OPUS_5_MODEL,
+    endpoint: "https://api.runinfra.ai/v1/chat/completions",
+    usageProvider: "runinfra",
+    keyName: "runinfra",
     wire: "chat.completions",
-    agentRunHosted: true
+    agentRunHosted: true,
+    upstreamModelId: PLATFORM_HOSTED_GLM_53_FLASH_MODEL
   },
   [PLATFORM_HOSTED_CLAUDE_FABLE_5_MODEL]: {
-    endpoint: "https://api.deepinfra.com/v1/openai/chat/completions",
-    usageProvider: "deepinfra",
-    keyName: "deepinfra",
-    upstreamModelId: PLATFORM_HOSTED_CLAUDE_FABLE_5_MODEL,
-    wire: "chat.completions",
-    agentRunHosted: true
-  },
-  [PLATFORM_HOSTED_CLAUDE_HAIKU_45_MODEL]: {
-    endpoint: "https://api.deepinfra.com/v1/openai/chat/completions",
-    usageProvider: "deepinfra",
-    keyName: "deepinfra",
-    upstreamModelId: PLATFORM_HOSTED_CLAUDE_HAIKU_45_MODEL,
-    wire: "chat.completions",
-    agentRunHosted: true
-  },
-  // GPT-6 / GPT-5.6 系（OpenAI 官方 chat.completions）：
-  // 平台价按短上下文价目定（长上下文档差价由平台吸收），见 platformHosted.ts。
-  [PLATFORM_HOSTED_GPT_6_ASTRA_MODEL]: {
-    endpoint: "https://api.openai.com/v1/chat/completions",
-    usageProvider: "openai",
-    keyName: "openai",
-    wire: "chat.completions",
-    agentRunHosted: true
-  },
-  [PLATFORM_HOSTED_GPT_56_SOL_MODEL]: {
-    endpoint: "https://api.openai.com/v1/chat/completions",
-    usageProvider: "openai",
-    keyName: "openai",
-    wire: "chat.completions",
-    agentRunHosted: true
-  },
-  [PLATFORM_HOSTED_GPT_56_TERRA_MODEL]: {
-    endpoint: "https://api.openai.com/v1/chat/completions",
-    usageProvider: "openai",
-    keyName: "openai",
-    wire: "chat.completions",
-    agentRunHosted: true
-  },
-  [PLATFORM_HOSTED_GPT_56_LUNA_MODEL]: {
-    endpoint: "https://api.openai.com/v1/chat/completions",
-    usageProvider: "openai",
-    keyName: "openai",
+    endpoint: "https://api.runinfra.ai/v1/chat/completions",
+    usageProvider: "runinfra",
+    keyName: "runinfra",
     wire: "chat.completions",
-    agentRunHosted: true
+    agentRunHosted: true,
+    upstreamModelId: PLATFORM_HOSTED_GLM_53_FLASH_MODEL
   },
   // Grok -> xAI（server agentRun 侧未分流，保持 no upstream route 报错）
   [PLATFORM_HOSTED_GROK_4_6_MODEL]: {
@@ -247,6 +199,15 @@ var PLATFORM_HOSTED_ROUTING_TABLE = {
     wire: "responses",
     agentRunHosted: true
   },
+  // Nemotron 标题 LLM 内部专用，走 chatProxy；agentRun 不分流。
+  [PLATFORM_HOSTED_NEMOTRON_35_LIGHTNING_MODEL]: {
+    endpoint: "https://api.runinfra.ai/v1/chat/completions",
+    usageProvider: "runinfra",
+    keyName: "runinfra",
+    upstreamModelId: "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16",
+    wire: "chat.completions",
+    agentRunHosted: false
+  },
   // DeepSeek Flash -> DeepSeek Responses API（agentRun 走专用 responses 编排）
   [PLATFORM_HOSTED_DEEPSEEK_FLASH_MODEL]: {
     endpoint: "https://api.deepseek.com/responses",
@@ -413,6 +374,14 @@ function buildClientVersionGateErrorDetail(decision) {
 }
 
 // packages/ai/llm/platformHosted.ts
+var PLATFORM_HOSTED_CLAUDE_MODELS = [
+  PLATFORM_HOSTED_CLAUDE_SONNET_5_MODEL,
+  PLATFORM_HOSTED_CLAUDE_OPUS_5_MODEL,
+  PLATFORM_HOSTED_CLAUDE_FABLE_5_MODEL
+];
+var isPlatformHostedClaudeModel = (model) => PLATFORM_HOSTED_CLAUDE_MODELS.some(
+  (m) => asTrimmedLowercaseString(model) === m
+);
 var PLATFORM_CREDITS_PER_USD = 8;
 var toPlatformCredits = (usdPerMillion) => Number((usdPerMillion * PLATFORM_CREDITS_PER_USD).toFixed(6));
 var CNY_UPSTREAM_MULTIPLIER = 1.2;
@@ -435,77 +404,18 @@ var PLATFORM_HOSTED_KIMI_PRICE = {
   // 24.0 credits (24 = 24)
 };
 var PLATFORM_HOSTED_KIMI_K3_PRICE = {
-  input: toPlatformCredits(3),
-  // 24.0 credits
-  inputCacheHit: toPlatformCredits(0.3),
-  // 2.4 credits
-  output: toPlatformCredits(15)
-  // 120.0 credits
+  input: 16,
+  inputCacheHit: 2,
+  output: 64
 };
 var PLATFORM_HOSTED_GLM_PRICE = {
   input: 7.84,
   // 11.2 credits
-  inputCacheHit: toPlatformCredits(0.14),
-  // 1.12 credits
+  inputCacheHit: toPlatformCredits(0.06),
+  // 0.48 credits
   output: 24.64
   // 35.2 credits (35.2 = 35.2)
 };
-var PLATFORM_HOSTED_CLAUDE_FABLE_5_PRICE = {
-  input: toPlatformCredits(10),
-  // 80 credits
-  output: toPlatformCredits(50)
-  // 400 credits
-};
-var PLATFORM_HOSTED_CLAUDE_OPUS_5_PRICE = {
-  input: toPlatformCredits(5),
-  // 40 credits
-  output: toPlatformCredits(25)
-  // 200 credits
-};
-var PLATFORM_HOSTED_CLAUDE_SONNET_5_PRICE = {
-  input: toPlatformCredits(3),
-  // 24 credits
-  output: toPlatformCredits(15)
-  // 120 credits
-};
-var PLATFORM_HOSTED_CLAUDE_HAIKU_45_PRICE = {
-  input: toPlatformCredits(1),
-  // 8 credits
-  output: toPlatformCredits(5)
-  // 40 credits
-};
-var PLATFORM_HOSTED_GPT_6_ASTRA_PRICE = {
-  input: toPlatformCredits(10),
-  // 80 credits
-  inputCacheHit: toPlatformCredits(1),
-  // 8 credits
-  output: toPlatformCredits(50)
-  // 400 credits
-};
-var PLATFORM_HOSTED_GPT_56_SOL_PRICE = {
-  input: toPlatformCredits(4),
-  // 32 credits
-  inputCacheHit: toPlatformCredits(0.4),
-  // 3.2 credits
-  output: toPlatformCredits(20)
-  // 160 credits
-};
-var PLATFORM_HOSTED_GPT_56_TERRA_PRICE = {
-  input: toPlatformCredits(2),
-  // 16 credits
-  inputCacheHit: toPlatformCredits(0.2),
-  // 1.6 credits
-  output: toPlatformCredits(12)
-  // 96 credits
-};
-var PLATFORM_HOSTED_GPT_56_LUNA_PRICE = {
-  input: toPlatformCredits(0.2),
-  // 1.6 credits
-  inputCacheHit: toPlatformCredits(0.02),
-  // 0.16 credits
-  output: toPlatformCredits(1.2)
-  // 9.6 credits
-};
 var PLATFORM_HOSTED_GLM_53_FLASH_PRICE = {
   input: toPlatformCredits(0.1),
   // 0.8 credits
@@ -629,6 +539,11 @@ var PLATFORM_HOSTED_DEEPSEEK_PRO_OFF_PEAK_PRICE = {
   output: toCnyCredits(13.5)
   // ¥13.5
 };
+var PLATFORM_HOSTED_NEMOTRON_35_LIGHTNING_PRICE = {
+  input: toPlatformCredits(0.05),
+  inputCacheHit: toPlatformCredits(0.01),
+  output: toPlatformCredits(0.15)
+};
 var PLATFORM_HOSTED_LEGACY_DEEPSEEK_MODELS = [
   "deepseek-v4-flash",
   "deepseek-v4-flash-vision-exp"
@@ -655,6 +570,16 @@ var isPlatformHostedDeepSeekModel = (model) => {
 var isPlatformHostedDeepseekModel = isPlatformHostedDeepSeekModel;
 var isPlatformHostedDeepSeekV4Model = isPlatformHostedDeepSeekModel;
 var platformHostedModels = [
+  {
+    name: PLATFORM_HOSTED_NEMOTRON_35_LIGHTNING_MODEL,
+    displayName: "Nemotron 3.5 Lightning 30B",
+    hasVision: false,
+    price: { ...PLATFORM_HOSTED_NEMOTRON_35_LIGHTNING_PRICE },
+    maxOutputTokens: 32768,
+    contextWindow: 262144,
+    supportsTool: true,
+    supportsReasoningEffort: true
+  },
   {
     name: PLATFORM_HOSTED_KIMI_K3_MODEL,
     displayName: "Kimi K3",
@@ -754,85 +679,9 @@ var platformHostedModels = [
     supportsReasoningEffort: true,
     minClientVersion: PLATFORM_HOSTED_GLM_53_FLASH_MIN_CLIENT_VERSION
   },
-  // Claude 系（真实 DeepInfra 模型，官方 id 与平台 id 同名）：无缓存价，
-  // 计费按 input 全价（见 calculatePrice 的 deepinfra 分支）。
-  {
-    name: PLATFORM_HOSTED_CLAUDE_SONNET_5_MODEL,
-    displayName: "Claude Sonnet 5",
-    hasVision: true,
-    price: { ...PLATFORM_HOSTED_CLAUDE_SONNET_5_PRICE },
-    maxOutputTokens: 4092,
-    contextWindow: 1e6,
-    supportsTool: false
-  },
-  {
-    name: PLATFORM_HOSTED_CLAUDE_OPUS_5_MODEL,
-    displayName: "Claude Opus 5",
-    hasVision: true,
-    price: { ...PLATFORM_HOSTED_CLAUDE_OPUS_5_PRICE },
-    maxOutputTokens: 4092,
-    contextWindow: 1e6,
-    supportsTool: false
-  },
-  {
-    name: PLATFORM_HOSTED_CLAUDE_FABLE_5_MODEL,
-    displayName: "Claude Fable 5",
-    hasVision: true,
-    price: { ...PLATFORM_HOSTED_CLAUDE_FABLE_5_PRICE },
-    maxOutputTokens: 4092,
-    contextWindow: 1e6,
-    supportsTool: false
-  },
-  {
-    name: PLATFORM_HOSTED_CLAUDE_HAIKU_45_MODEL,
-    displayName: "Claude Haiku 4.5",
-    hasVision: true,
-    price: { ...PLATFORM_HOSTED_CLAUDE_HAIKU_45_PRICE },
-    maxOutputTokens: 4092,
-    contextWindow: 2e5,
-    supportsTool: false
-  },
-  // GPT-6 / GPT-5.6 系：OpenAI 官方 chat.completions，平台价按短上下文价目。
-  {
-    name: PLATFORM_HOSTED_GPT_6_ASTRA_MODEL,
-    displayName: "GPT-6 Astra",
-    hasVision: true,
-    price: { ...PLATFORM_HOSTED_GPT_6_ASTRA_PRICE },
-    maxOutputTokens: 128e3,
-    contextWindow: 105e4,
-    supportsTool: true,
-    supportsReasoningEffort: true
-  },
-  {
-    name: PLATFORM_HOSTED_GPT_56_SOL_MODEL,
-    displayName: "GPT-5.6 Sol",
-    hasVision: true,
-    price: { ...PLATFORM_HOSTED_GPT_56_SOL_PRICE },
-    maxOutputTokens: 128e3,
-    contextWindow: 105e4,
-    supportsTool: true,
-    supportsReasoningEffort: true
-  },
-  {
-    name: PLATFORM_HOSTED_GPT_56_TERRA_MODEL,
-    displayName: "GPT-5.6 Terra",
-    hasVision: true,
-    price: { ...PLATFORM_HOSTED_GPT_56_TERRA_PRICE },
-    maxOutputTokens: 128e3,
-    contextWindow: 105e4,
-    supportsTool: true,
-    supportsReasoningEffort: true
-  },
-  {
-    name: PLATFORM_HOSTED_GPT_56_LUNA_MODEL,
-    displayName: "GPT-5.6 Luna",
-    hasVision: true,
-    price: { ...PLATFORM_HOSTED_GPT_56_LUNA_PRICE },
-    maxOutputTokens: 128e3,
-    contextWindow: 105e4,
-    supportsTool: true,
-    supportsReasoningEffort: true
-  },
+  // Claude 系已下架（2026-09-01），不再出现在 nolo 托管模型列表；旧模型名请求由
+  // platformHostedRoutingTable 重映射到 glm-5-3-flash 做兼容。deepinfra 直连通道
+  // （用户自有 key）不受影响，见 ./deepinfra.ts。
   {
     name: PLATFORM_HOSTED_GROK_4_6_MODEL,
     displayName: "Grok 4.6",
@@ -1000,6 +849,7 @@ export {
   isFireworksKimiModel,
   isDeepInfraKimiModel,
   resolveFireworksKimiModel,
+  PLATFORM_HOSTED_GLM_53_FLASH_MODEL,
   PLATFORM_HOSTED_DEEPSEEK_FLASH_MODEL,
   PLATFORM_HOSTED_DEEPSEEK_PRO_MODEL,
   resolvePlatformHostedRouting,
@@ -1008,6 +858,7 @@ export {
   buildClientVersionUpgradeMessage,
   evaluatePlatformHostedClientVersionGate,
   buildClientVersionGateErrorDetail,
+  isPlatformHostedClaudeModel,
   toCnyCredits,
   PLATFORM_HOSTED_GLM_53_FLASH_BASETEN_PRICE,
   isPlatformHostedOpenAIImageModel,

package/{chunk-NG4UT4K4.js → chunk-6IT5ECCC.js}

@@ -4,7 +4,7 @@ import {
   openrouterModels,
   platformHostedModels,
   zaiModels
-} from "./chunk-AJHEFRGJ.js";
+} from "./chunk-5YIL5WDC.js";
 
 // packages/ai/llm/modelAbility.ts
 var MODEL_ABILITY_TABLE = {