@omnicross/contracts 0.1.7 → 0.1.9

package/dist/canonical-models.cjs

@@ -29,9 +29,9 @@ __export(canonical_models_exports, {
 });
 module.exports = __toCommonJS(canonical_models_exports);
 var OPENAI_MODELS = {
-  "gpt-5.6-sol": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
-  "gpt-5.6-terra": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
-  "gpt-5.6-luna": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-sol": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-terra": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-luna": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.5": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.4": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.4-mini": { category: "reasoning", contextLength: 4e5, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
@@ -200,6 +200,7 @@ function normalizeModelId(rawId) {
   return noSuffix.toLowerCase();
 }
 var MODEL_ALIASES = {
+  "gpt-5.6": "gpt-5.6-sol",
   "deepseek-chat": "deepseek-v3",
   "deepseek-reasoner": "deepseek-r1"
 };

package/dist/canonical-models.d.cts

@@ -71,9 +71,8 @@ declare function normalizeModelId(rawId: string): string;
  * assertion at module load enforces this so the alias map can't drift to
  * point at non-existent entries.
  *
- * Currently covers DeepSeek's marketing names (`deepseek-chat` is the
- * marketing alias for the latest non-reasoning model — historically v3,
- * may rotate as new releases land).
+ * Includes official API aliases plus vendor marketing names. DeepSeek's
+ * `deepseek-chat` historically targets v3 and may rotate as new releases land.
  */
 declare const MODEL_ALIASES: Record<string, string>;
 /**

package/dist/canonical-models.d.ts

@@ -71,9 +71,8 @@ declare function normalizeModelId(rawId: string): string;
  * assertion at module load enforces this so the alias map can't drift to
  * point at non-existent entries.
  *
- * Currently covers DeepSeek's marketing names (`deepseek-chat` is the
- * marketing alias for the latest non-reasoning model — historically v3,
- * may rotate as new releases land).
+ * Includes official API aliases plus vendor marketing names. DeepSeek's
+ * `deepseek-chat` historically targets v3 and may rotate as new releases land.
  */
 declare const MODEL_ALIASES: Record<string, string>;
 /**

package/dist/canonical-models.js

@@ -1,8 +1,8 @@
 // src/canonical-models.ts
 var OPENAI_MODELS = {
-  "gpt-5.6-sol": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
-  "gpt-5.6-terra": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
-  "gpt-5.6-luna": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-sol": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-terra": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-luna": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.5": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.4": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.4-mini": { category: "reasoning", contextLength: 4e5, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
@@ -171,6 +171,7 @@ function normalizeModelId(rawId) {
   return noSuffix.toLowerCase();
 }
 var MODEL_ALIASES = {
+  "gpt-5.6": "gpt-5.6-sol",
   "deepseek-chat": "deepseek-v3",
   "deepseek-reasoner": "deepseek-r1"
 };

package/dist/index.cjs

@@ -35,6 +35,7 @@ __export(index_exports, {
   EFFORT_RATIO: () => EFFORT_RATIO,
   EXTENDED_CONTEXT_CAPABLE_MODELS: () => EXTENDED_CONTEXT_CAPABLE_MODELS,
   KNOWN_MODELS: () => KNOWN_MODELS,
+  LEGACY_THINKING_TOKEN_LIMITS: () => LEGACY_THINKING_TOKEN_LIMITS,
   LLM_PROVIDER_PRESETS: () => LLM_PROVIDER_PRESETS,
   MAX_CONCURRENCY_DEFAULTS: () => MAX_CONCURRENCY_DEFAULTS,
   MODEL_ALIASES: () => MODEL_ALIASES,
@@ -51,6 +52,7 @@ __export(index_exports, {
   buildQwenThinkingConfig: () => buildQwenThinkingConfig,
   calculateThinkingBudget: () => calculateThinkingBudget,
   canDisableThinking: () => canDisableThinking,
+  findLegacyTokenLimit: () => findLegacyTokenLimit,
   findTokenLimit: () => findTokenLimit,
   getAllProviderPresets: () => getAllProviderPresets,
   getClaudeMaxTokens: () => getClaudeMaxTokens,
@@ -92,9 +94,9 @@ var DEFAULT_BILLING_CONFIG = {
 
 // src/canonical-models.ts
 var OPENAI_MODELS = {
-  "gpt-5.6-sol": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
-  "gpt-5.6-terra": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
-  "gpt-5.6-luna": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-sol": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-terra": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-luna": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.5": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.4": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.4-mini": { category: "reasoning", contextLength: 4e5, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
@@ -263,6 +265,7 @@ function normalizeModelId(rawId) {
   return noSuffix.toLowerCase();
 }
 var MODEL_ALIASES = {
+  "gpt-5.6": "gpt-5.6-sol",
   "deepseek-chat": "deepseek-v3",
   "deepseek-reasoner": "deepseek-r1"
 };
@@ -573,6 +576,7 @@ var DEFAULT_SEED_PRESET_IDS = [
   // MiniMax + Xiaomi MiMo bumped up next to Kimi per UX preference —
   // popular Anthropic-format Chinese providers cluster together.
   "openrouter",
+  "openrouter-response",
   "zhipu",
   "zhipu-bigmodel",
   "kimi",
@@ -1615,6 +1619,63 @@ var openrouter_default = {
   ]
 };
 
+// src/provider-presets/presets/openrouter-response.json
+var openrouter_response_default = {
+  id: "openrouter-response",
+  presetId: "openrouter-response",
+  name: "OpenRouter (Responses API)",
+  apiFormat: "openai-response",
+  api_base_url: "https://openrouter.ai/api/v1",
+  modelsEndpoint: "https://openrouter.ai/api/v1/models",
+  models: [
+    "anthropic/claude-sonnet-4-6",
+    "openai/gpt-5",
+    "google/gemini-3.1-pro",
+    "meta-llama/llama-3.3-70b-instruct",
+    "deepseek/deepseek-v3.2"
+  ],
+  modelConfigs: [
+    {
+      id: "anthropic/claude-sonnet-4-6",
+      name: "Claude Sonnet 4.6 (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "openai/gpt-5",
+      name: "GPT-5 (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "google/gemini-3.1-pro",
+      name: "Gemini 3.1 Pro (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "meta-llama/llama-3.3-70b-instruct",
+      name: "Llama 3.3 70B (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "deepseek/deepseek-v3.2",
+      name: "DeepSeek V3.2 (OpenRouter Responses)",
+      enabled: true
+    }
+  ],
+  description: "OpenRouter Responses API \u2014 \u539F\u6837\u900F\u4F20 Responses \u53C2\u6570\uFF0C\u652F\u6301 Claude \u663E\u5F0F\u7F13\u5B58\u63A7\u5236",
+  icon: "openrouter",
+  website: "https://openrouter.ai",
+  docsUrl: "https://openrouter.ai/docs/api-reference/responses/overview",
+  defaultSettings: {
+    temperature: 0.7,
+    maxTokens: 4096
+  },
+  isSystem: true,
+  features: [
+    "vision",
+    "reasoning"
+  ]
+};
+
 // src/provider-presets/presets/perplexity.json
 var perplexity_default = {
   id: "perplexity",
@@ -2335,6 +2396,7 @@ var RAW_PRESETS_IN_ORDER = [
   openai_response_default,
   siliconflow_default,
   openrouter_default,
+  openrouter_response_default,
   mistral_default,
   together_default,
   perplexity_default,
@@ -2524,7 +2586,7 @@ var EFFORT_RATIO = {
   max: 0.95
 };
 var DEFAULT_MAX_TOKENS = 4096;
-var THINKING_TOKEN_MAP = {
+var LEGACY_THINKING_TOKEN_LIMITS = {
   // Gemini
   "gemini-2\\.5-flash-lite": { min: 512, max: 24576 },
   "gemini-.*-flash": { min: 0, max: 24576 },
@@ -2580,6 +2642,7 @@ var THINKING_TOKEN_MAP = {
   // Doubao
   "doubao.*think": { min: 0, max: 16384 }
 };
+var THINKING_TOKEN_MAP = LEGACY_THINKING_TOKEN_LIMITS;
 var REASONING_MODEL_PATTERNS = [
   /^o[134](-mini|-preview|-pro)?$/i,
   /^gpt-5(\.\d)?(-pro|-codex|-codex-max)?/i,
@@ -2612,8 +2675,14 @@ function findTokenLimit(modelOrId) {
     if (modelOrId.thinkingTokenLimit) return modelOrId.thinkingTokenLimit;
     return findTokenLimit(modelOrId.id);
   }
-  const lowerModelId = modelOrId.toLowerCase();
-  for (const [pattern, limit] of Object.entries(THINKING_TOKEN_MAP)) {
+  const canonicalLimit = lookupCanonicalCapabilities(modelOrId)?.thinkingTokenLimit;
+  if (canonicalLimit) return canonicalLimit;
+  return findLegacyTokenLimit(modelOrId);
+}
+function findLegacyTokenLimit(modelId) {
+  if (!modelId) return null;
+  const lowerModelId = modelId.toLowerCase();
+  for (const [pattern, limit] of Object.entries(LEGACY_THINKING_TOKEN_LIMITS)) {
     const regex = new RegExp(pattern, "i");
     if (regex.test(lowerModelId)) {
       return limit;
@@ -2743,6 +2812,7 @@ function isLocalProvider(id) {
   EFFORT_RATIO,
   EXTENDED_CONTEXT_CAPABLE_MODELS,
   KNOWN_MODELS,
+  LEGACY_THINKING_TOKEN_LIMITS,
   LLM_PROVIDER_PRESETS,
   MAX_CONCURRENCY_DEFAULTS,
   MODEL_ALIASES,
@@ -2759,6 +2829,7 @@ function isLocalProvider(id) {
   buildQwenThinkingConfig,
   calculateThinkingBudget,
   canDisableThinking,
+  findLegacyTokenLimit,
   findTokenLimit,
   getAllProviderPresets,
   getClaudeMaxTokens,

package/dist/index.d.cts

@@ -15,7 +15,7 @@ export { MessageBlock, MessageBlockBase, MessageBlockType, TextBlock, ThinkingBl
 export { DEFAULT_LITELLM_PRICING_URL, DEFAULT_OPENROUTER_PRICING_URL, PricingConflict, PricingConflictDecision, PricingEntry, PricingEntryInput, PricingFetchResult, PricingResolution, PricingSource, PricingSourceRefreshResult } from './pricing-types.cjs';
 export { CATALOG_VERSION, CODING_PLAN_URL_PRESETS, DEFAULT_SEED_PRESET_IDS, LLM_PROVIDER_PRESETS, MAX_CONCURRENCY_DEFAULTS, PROVIDER_MODEL_MAPPINGS, PROVIDER_SEARCH_CONFIGS, getAllProviderPresets, getCodingPlanBaseUrl, getPresetById, getPresetRevision, getProviderSearchConfig, resolveFollowProviderModel } from './provider-presets/index.cjs';
 export { SUBSCRIPTION_MODEL_CATALOG, subscriptionProviderHasCatalog } from './subscription-model-catalog.cjs';
-export { CANNOT_DISABLE_THINKING_PATTERNS, DEFAULT_MAX_TOKENS, EFFORT_RATIO, REASONING_MODEL_PATTERNS, THINKING_TOKEN_MAP, buildAnthropicThinking, buildGeminiThinkingConfig, buildQwenThinkingConfig, calculateThinkingBudget, canDisableThinking, findTokenLimit, getClaudeMaxTokens, getOpenAIReasoningEffort, isReasoningModel } from './thinking-config.cjs';
+export { CANNOT_DISABLE_THINKING_PATTERNS, DEFAULT_MAX_TOKENS, EFFORT_RATIO, LEGACY_THINKING_TOKEN_LIMITS, REASONING_MODEL_PATTERNS, THINKING_TOKEN_MAP, buildAnthropicThinking, buildGeminiThinkingConfig, buildQwenThinkingConfig, calculateThinkingBudget, canDisableThinking, findLegacyTokenLimit, findTokenLimit, getClaudeMaxTokens, getOpenAIReasoningEffort, isReasoningModel } from './thinking-config.cjs';
 export { ApiKeyUsageRow, MessageUsageRow, ModelUsageRow, SessionCacheStats, UsageCacheKeySource, UsageDateRange, UsageEventInput, UsageEventRecord, UsageQueryParams, UsageTimeBucket, UsageTimeSeriesBucket, UsageTotals } from './usage-stats-types.cjs';
 export { UsageEngineOrigin, UsageTokens } from './usage-types.cjs';
 export { DEFAULT_VOUCHER_CONFIG, VoucherConfig, VoucherCreated, VoucherGrant, VoucherInfo, VoucherRecord, VoucherRedeemResult, VoucherStatus, VoucherType } from './voucher-types.cjs';

package/dist/index.d.ts

@@ -15,7 +15,7 @@ export { MessageBlock, MessageBlockBase, MessageBlockType, TextBlock, ThinkingBl
 export { DEFAULT_LITELLM_PRICING_URL, DEFAULT_OPENROUTER_PRICING_URL, PricingConflict, PricingConflictDecision, PricingEntry, PricingEntryInput, PricingFetchResult, PricingResolution, PricingSource, PricingSourceRefreshResult } from './pricing-types.js';
 export { CATALOG_VERSION, CODING_PLAN_URL_PRESETS, DEFAULT_SEED_PRESET_IDS, LLM_PROVIDER_PRESETS, MAX_CONCURRENCY_DEFAULTS, PROVIDER_MODEL_MAPPINGS, PROVIDER_SEARCH_CONFIGS, getAllProviderPresets, getCodingPlanBaseUrl, getPresetById, getPresetRevision, getProviderSearchConfig, resolveFollowProviderModel } from './provider-presets/index.js';
 export { SUBSCRIPTION_MODEL_CATALOG, subscriptionProviderHasCatalog } from './subscription-model-catalog.js';
-export { CANNOT_DISABLE_THINKING_PATTERNS, DEFAULT_MAX_TOKENS, EFFORT_RATIO, REASONING_MODEL_PATTERNS, THINKING_TOKEN_MAP, buildAnthropicThinking, buildGeminiThinkingConfig, buildQwenThinkingConfig, calculateThinkingBudget, canDisableThinking, findTokenLimit, getClaudeMaxTokens, getOpenAIReasoningEffort, isReasoningModel } from './thinking-config.js';
+export { CANNOT_DISABLE_THINKING_PATTERNS, DEFAULT_MAX_TOKENS, EFFORT_RATIO, LEGACY_THINKING_TOKEN_LIMITS, REASONING_MODEL_PATTERNS, THINKING_TOKEN_MAP, buildAnthropicThinking, buildGeminiThinkingConfig, buildQwenThinkingConfig, calculateThinkingBudget, canDisableThinking, findLegacyTokenLimit, findTokenLimit, getClaudeMaxTokens, getOpenAIReasoningEffort, isReasoningModel } from './thinking-config.js';
 export { ApiKeyUsageRow, MessageUsageRow, ModelUsageRow, SessionCacheStats, UsageCacheKeySource, UsageDateRange, UsageEventInput, UsageEventRecord, UsageQueryParams, UsageTimeBucket, UsageTimeSeriesBucket, UsageTotals } from './usage-stats-types.js';
 export { UsageEngineOrigin, UsageTokens } from './usage-types.js';
 export { DEFAULT_VOUCHER_CONFIG, VoucherConfig, VoucherCreated, VoucherGrant, VoucherInfo, VoucherRecord, VoucherRedeemResult, VoucherStatus, VoucherType } from './voucher-types.js';

package/dist/index.js

@@ -15,9 +15,9 @@ var DEFAULT_BILLING_CONFIG = {
 
 // src/canonical-models.ts
 var OPENAI_MODELS = {
-  "gpt-5.6-sol": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
-  "gpt-5.6-terra": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
-  "gpt-5.6-luna": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-sol": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-terra": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.6-luna": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.5": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.4": { category: "reasoning", contextLength: 105e4, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
   "gpt-5.4-mini": { category: "reasoning", contextLength: 4e5, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
@@ -186,6 +186,7 @@ function normalizeModelId(rawId) {
   return noSuffix.toLowerCase();
 }
 var MODEL_ALIASES = {
+  "gpt-5.6": "gpt-5.6-sol",
   "deepseek-chat": "deepseek-v3",
   "deepseek-reasoner": "deepseek-r1"
 };
@@ -496,6 +497,7 @@ var DEFAULT_SEED_PRESET_IDS = [
   // MiniMax + Xiaomi MiMo bumped up next to Kimi per UX preference —
   // popular Anthropic-format Chinese providers cluster together.
   "openrouter",
+  "openrouter-response",
   "zhipu",
   "zhipu-bigmodel",
   "kimi",
@@ -1538,6 +1540,63 @@ var openrouter_default = {
   ]
 };
 
+// src/provider-presets/presets/openrouter-response.json
+var openrouter_response_default = {
+  id: "openrouter-response",
+  presetId: "openrouter-response",
+  name: "OpenRouter (Responses API)",
+  apiFormat: "openai-response",
+  api_base_url: "https://openrouter.ai/api/v1",
+  modelsEndpoint: "https://openrouter.ai/api/v1/models",
+  models: [
+    "anthropic/claude-sonnet-4-6",
+    "openai/gpt-5",
+    "google/gemini-3.1-pro",
+    "meta-llama/llama-3.3-70b-instruct",
+    "deepseek/deepseek-v3.2"
+  ],
+  modelConfigs: [
+    {
+      id: "anthropic/claude-sonnet-4-6",
+      name: "Claude Sonnet 4.6 (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "openai/gpt-5",
+      name: "GPT-5 (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "google/gemini-3.1-pro",
+      name: "Gemini 3.1 Pro (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "meta-llama/llama-3.3-70b-instruct",
+      name: "Llama 3.3 70B (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "deepseek/deepseek-v3.2",
+      name: "DeepSeek V3.2 (OpenRouter Responses)",
+      enabled: true
+    }
+  ],
+  description: "OpenRouter Responses API \u2014 \u539F\u6837\u900F\u4F20 Responses \u53C2\u6570\uFF0C\u652F\u6301 Claude \u663E\u5F0F\u7F13\u5B58\u63A7\u5236",
+  icon: "openrouter",
+  website: "https://openrouter.ai",
+  docsUrl: "https://openrouter.ai/docs/api-reference/responses/overview",
+  defaultSettings: {
+    temperature: 0.7,
+    maxTokens: 4096
+  },
+  isSystem: true,
+  features: [
+    "vision",
+    "reasoning"
+  ]
+};
+
 // src/provider-presets/presets/perplexity.json
 var perplexity_default = {
   id: "perplexity",
@@ -2258,6 +2317,7 @@ var RAW_PRESETS_IN_ORDER = [
   openai_response_default,
   siliconflow_default,
   openrouter_default,
+  openrouter_response_default,
   mistral_default,
   together_default,
   perplexity_default,
@@ -2447,7 +2507,7 @@ var EFFORT_RATIO = {
   max: 0.95
 };
 var DEFAULT_MAX_TOKENS = 4096;
-var THINKING_TOKEN_MAP = {
+var LEGACY_THINKING_TOKEN_LIMITS = {
   // Gemini
   "gemini-2\\.5-flash-lite": { min: 512, max: 24576 },
   "gemini-.*-flash": { min: 0, max: 24576 },
@@ -2503,6 +2563,7 @@ var THINKING_TOKEN_MAP = {
   // Doubao
   "doubao.*think": { min: 0, max: 16384 }
 };
+var THINKING_TOKEN_MAP = LEGACY_THINKING_TOKEN_LIMITS;
 var REASONING_MODEL_PATTERNS = [
   /^o[134](-mini|-preview|-pro)?$/i,
   /^gpt-5(\.\d)?(-pro|-codex|-codex-max)?/i,
@@ -2535,8 +2596,14 @@ function findTokenLimit(modelOrId) {
     if (modelOrId.thinkingTokenLimit) return modelOrId.thinkingTokenLimit;
     return findTokenLimit(modelOrId.id);
   }
-  const lowerModelId = modelOrId.toLowerCase();
-  for (const [pattern, limit] of Object.entries(THINKING_TOKEN_MAP)) {
+  const canonicalLimit = lookupCanonicalCapabilities(modelOrId)?.thinkingTokenLimit;
+  if (canonicalLimit) return canonicalLimit;
+  return findLegacyTokenLimit(modelOrId);
+}
+function findLegacyTokenLimit(modelId) {
+  if (!modelId) return null;
+  const lowerModelId = modelId.toLowerCase();
+  for (const [pattern, limit] of Object.entries(LEGACY_THINKING_TOKEN_LIMITS)) {
     const regex = new RegExp(pattern, "i");
     if (regex.test(lowerModelId)) {
       return limit;
@@ -2665,6 +2732,7 @@ export {
   EFFORT_RATIO,
   EXTENDED_CONTEXT_CAPABLE_MODELS,
   KNOWN_MODELS,
+  LEGACY_THINKING_TOKEN_LIMITS,
   LLM_PROVIDER_PRESETS,
   MAX_CONCURRENCY_DEFAULTS,
   MODEL_ALIASES,
@@ -2681,6 +2749,7 @@ export {
   buildQwenThinkingConfig,
   calculateThinkingBudget,
   canDisableThinking,
+  findLegacyTokenLimit,
   findTokenLimit,
   getAllProviderPresets,
   getClaudeMaxTokens,

package/dist/provider-presets/index.cjs

@@ -232,6 +232,7 @@ var DEFAULT_SEED_PRESET_IDS = [
   // MiniMax + Xiaomi MiMo bumped up next to Kimi per UX preference —
   // popular Anthropic-format Chinese providers cluster together.
   "openrouter",
+  "openrouter-response",
   "zhipu",
   "zhipu-bigmodel",
   "kimi",
@@ -1274,6 +1275,63 @@ var openrouter_default = {
   ]
 };
 
+// src/provider-presets/presets/openrouter-response.json
+var openrouter_response_default = {
+  id: "openrouter-response",
+  presetId: "openrouter-response",
+  name: "OpenRouter (Responses API)",
+  apiFormat: "openai-response",
+  api_base_url: "https://openrouter.ai/api/v1",
+  modelsEndpoint: "https://openrouter.ai/api/v1/models",
+  models: [
+    "anthropic/claude-sonnet-4-6",
+    "openai/gpt-5",
+    "google/gemini-3.1-pro",
+    "meta-llama/llama-3.3-70b-instruct",
+    "deepseek/deepseek-v3.2"
+  ],
+  modelConfigs: [
+    {
+      id: "anthropic/claude-sonnet-4-6",
+      name: "Claude Sonnet 4.6 (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "openai/gpt-5",
+      name: "GPT-5 (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "google/gemini-3.1-pro",
+      name: "Gemini 3.1 Pro (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "meta-llama/llama-3.3-70b-instruct",
+      name: "Llama 3.3 70B (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "deepseek/deepseek-v3.2",
+      name: "DeepSeek V3.2 (OpenRouter Responses)",
+      enabled: true
+    }
+  ],
+  description: "OpenRouter Responses API \u2014 \u539F\u6837\u900F\u4F20 Responses \u53C2\u6570\uFF0C\u652F\u6301 Claude \u663E\u5F0F\u7F13\u5B58\u63A7\u5236",
+  icon: "openrouter",
+  website: "https://openrouter.ai",
+  docsUrl: "https://openrouter.ai/docs/api-reference/responses/overview",
+  defaultSettings: {
+    temperature: 0.7,
+    maxTokens: 4096
+  },
+  isSystem: true,
+  features: [
+    "vision",
+    "reasoning"
+  ]
+};
+
 // src/provider-presets/presets/perplexity.json
 var perplexity_default = {
   id: "perplexity",
@@ -1994,6 +2052,7 @@ var RAW_PRESETS_IN_ORDER = [
   openai_response_default,
   siliconflow_default,
   openrouter_default,
+  openrouter_response_default,
   mistral_default,
   together_default,
   perplexity_default,

package/dist/provider-presets/index.js

@@ -193,6 +193,7 @@ var DEFAULT_SEED_PRESET_IDS = [
   // MiniMax + Xiaomi MiMo bumped up next to Kimi per UX preference —
   // popular Anthropic-format Chinese providers cluster together.
   "openrouter",
+  "openrouter-response",
   "zhipu",
   "zhipu-bigmodel",
   "kimi",
@@ -1235,6 +1236,63 @@ var openrouter_default = {
   ]
 };
 
+// src/provider-presets/presets/openrouter-response.json
+var openrouter_response_default = {
+  id: "openrouter-response",
+  presetId: "openrouter-response",
+  name: "OpenRouter (Responses API)",
+  apiFormat: "openai-response",
+  api_base_url: "https://openrouter.ai/api/v1",
+  modelsEndpoint: "https://openrouter.ai/api/v1/models",
+  models: [
+    "anthropic/claude-sonnet-4-6",
+    "openai/gpt-5",
+    "google/gemini-3.1-pro",
+    "meta-llama/llama-3.3-70b-instruct",
+    "deepseek/deepseek-v3.2"
+  ],
+  modelConfigs: [
+    {
+      id: "anthropic/claude-sonnet-4-6",
+      name: "Claude Sonnet 4.6 (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "openai/gpt-5",
+      name: "GPT-5 (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "google/gemini-3.1-pro",
+      name: "Gemini 3.1 Pro (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "meta-llama/llama-3.3-70b-instruct",
+      name: "Llama 3.3 70B (OpenRouter Responses)",
+      enabled: true
+    },
+    {
+      id: "deepseek/deepseek-v3.2",
+      name: "DeepSeek V3.2 (OpenRouter Responses)",
+      enabled: true
+    }
+  ],
+  description: "OpenRouter Responses API \u2014 \u539F\u6837\u900F\u4F20 Responses \u53C2\u6570\uFF0C\u652F\u6301 Claude \u663E\u5F0F\u7F13\u5B58\u63A7\u5236",
+  icon: "openrouter",
+  website: "https://openrouter.ai",
+  docsUrl: "https://openrouter.ai/docs/api-reference/responses/overview",
+  defaultSettings: {
+    temperature: 0.7,
+    maxTokens: 4096
+  },
+  isSystem: true,
+  features: [
+    "vision",
+    "reasoning"
+  ]
+};
+
 // src/provider-presets/presets/perplexity.json
 var perplexity_default = {
   id: "perplexity",
@@ -1955,6 +2013,7 @@ var RAW_PRESETS_IN_ORDER = [
   openai_response_default,
   siliconflow_default,
   openrouter_default,
+  openrouter_response_default,
   mistral_default,
   together_default,
   perplexity_default,