@omnicross/contracts 0.1.7 → 0.1.9

package/dist/thinking-config.js

@@ -1,3 +1,205 @@
+// 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", "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 } },
+  "gpt-5.3-codex": { category: "code", contextLength: 4e5, maxTokens: 128e3, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 128e3 } },
+  "gpt-5.2": { category: "reasoning", contextLength: 105e4, maxTokens: 65536, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "minimal", "low", "medium", "high", "xhigh"], thinkingTokenLimit: { min: 0, max: 65536 } },
+  "gpt-5": { category: "reasoning", contextLength: 105e4, maxTokens: 65536, reasoning: true, vision: true, functionCall: true },
+  "gpt-4.1": { category: "chat", contextLength: 105e4, maxTokens: 32768, vision: true, functionCall: true },
+  "gpt-4o": { category: "chat", contextLength: 128e3, maxTokens: 16384, vision: true, functionCall: true },
+  "gpt-4o-mini": { category: "chat", contextLength: 128e3, maxTokens: 16384, functionCall: true },
+  "o3": { category: "reasoning", contextLength: 2e5, maxTokens: 1e5, reasoning: true, thinkingLevels: ["low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 65536 } },
+  "o4-mini": { category: "reasoning", contextLength: 2e5, maxTokens: 1e5, reasoning: true, thinkingLevels: ["low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } }
+};
+var ANTHROPIC_MODELS = {
+  "claude-fable-5": { category: "reasoning", contextLength: 2e5, maxTokens: 128e3, vision: true, functionCall: true, reasoning: true },
+  "claude-opus-5": { category: "reasoning", contextLength: 2e5, maxTokens: 128e3, vision: true, functionCall: true, reasoning: true },
+  "claude-opus-4-8": { category: "reasoning", contextLength: 2e5, maxTokens: 128e3, vision: true, functionCall: true, reasoning: true },
+  "claude-sonnet-5": { category: "chat", contextLength: 2e5, maxTokens: 128e3, vision: true, functionCall: true, reasoning: true },
+  "claude-opus-4-7": { category: "reasoning", contextLength: 2e5, maxTokens: 128e3, vision: true, functionCall: true, reasoning: true },
+  "claude-opus-4-7[1m]": { category: "reasoning", contextLength: 1e6, maxTokens: 128e3, vision: true, functionCall: true, reasoning: true },
+  "claude-opus-4-6": { category: "reasoning", contextLength: 2e5, maxTokens: 128e3, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 1024, max: 128e3 } },
+  "claude-opus-4-6[1m]": { category: "reasoning", contextLength: 1e6, maxTokens: 128e3, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 1024, max: 128e3 } },
+  "claude-sonnet-4-6": { category: "chat", contextLength: 2e5, maxTokens: 64e3, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 1024, max: 64e3 } },
+  "claude-sonnet-4-6[1m]": { category: "chat", contextLength: 1e6, maxTokens: 64e3, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 1024, max: 64e3 } },
+  "claude-haiku-4-5": { category: "chat", contextLength: 2e5, maxTokens: 64e3, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high", "xhigh", "max"], thinkingTokenLimit: { min: 1024, max: 64e3 } }
+};
+var GEMINI_MODELS = {
+  "gemini-3.5-flash": { category: "chat", contextLength: 1e6, maxTokens: 65536, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 24576 } },
+  "gemini-3-flash": { category: "chat", contextLength: 1e6, maxTokens: 65536, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 24576 } },
+  "gemini-3.1-pro": { category: "chat", contextLength: 1e6, maxTokens: 65536, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["low", "medium", "high"], thinkingTokenLimit: { min: 128, max: 32768 } }
+};
+var GROK_MODELS = {
+  "grok-4.5": { category: "chat", contextLength: 5e5, maxTokens: 32768, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } },
+  "grok-4.3": { category: "chat", contextLength: 1e6, maxTokens: 32768, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } },
+  "grok-4.20": { category: "chat", contextLength: 131072, maxTokens: 32768, vision: true, functionCall: true },
+  "grok-4.20-reasoning": { category: "reasoning", contextLength: 131072, maxTokens: 32768, reasoning: true, vision: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } },
+  "grok-4.20-multi-agent": { category: "reasoning", contextLength: 131072, maxTokens: 32768, reasoning: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } }
+};
+var DEEPSEEK_MODELS = {
+  "deepseek-v4-flash": { category: "chat", contextLength: 1e6, maxTokens: 384e3, functionCall: true, reasoning: true, thinkingLevels: ["none", "high", "max"], thinkingTokenLimit: { min: 0, max: 32768 } },
+  "deepseek-v4-pro": { category: "reasoning", contextLength: 1e6, maxTokens: 384e3, functionCall: true, reasoning: true, thinkingLevels: ["none", "high", "max"], thinkingTokenLimit: { min: 0, max: 32768 } },
+  "deepseek-v3": { category: "chat", contextLength: 65536, maxTokens: 8192, functionCall: true },
+  "deepseek-v3.2": { category: "code", contextLength: 256e3, maxTokens: 32768, functionCall: true, reasoning: true },
+  "deepseek-r1": { category: "reasoning", contextLength: 64e3, maxTokens: 8192, reasoning: true, thinkingTokenLimit: { min: 0, max: 32768 } }
+};
+var ZHIPU_MODELS = {
+  "glm-5.2": { category: "chat", contextLength: 1048576, maxTokens: 131072, functionCall: true },
+  "glm-5": { category: "chat", contextLength: 2e5, maxTokens: 128e3, functionCall: true },
+  "glm-5.1": { category: "chat", contextLength: 2e5, maxTokens: 128e3, functionCall: true },
+  "glm-4.5-air": { category: "chat", contextLength: 128e3, maxTokens: 32768, functionCall: true },
+  "glm-4.6v": { category: "chat", contextLength: 128e3, maxTokens: 32768, vision: true, functionCall: true },
+  "glm-4.7": { category: "code", contextLength: 2e5, maxTokens: 128e3, functionCall: true }
+};
+var KIMI_MODELS = {
+  "kimi-k3": { category: "code", contextLength: 1048576, maxTokens: 1048576, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } },
+  "kimi-k2.6": { category: "code", contextLength: 256e3, maxTokens: 32768, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } },
+  "kimi-k2.5": { category: "chat", contextLength: 256e3, maxTokens: 65535, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } },
+  "kimi-k2-thinking": { category: "reasoning", contextLength: 256e3, reasoning: true, functionCall: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } }
+};
+var DASHSCOPE_QWEN_MODELS = {
+  "qwen3-max": { category: "chat", contextLength: 262144, maxTokens: 65536, functionCall: true, webSearch: true },
+  "qwen3.5-plus": { category: "chat", contextLength: 1e6, vision: true, functionCall: true, webSearch: true },
+  "qwen-turbo": { category: "chat", contextLength: 1e6, vision: true, webSearch: true },
+  "qwen3-coder-plus": { category: "code", contextLength: 1e6, functionCall: true },
+  "qwen-3-coder-480b": { category: "code", contextLength: 131072, maxTokens: 8192, functionCall: true },
+  "qwen-3-32b": { category: "chat", contextLength: 131072, maxTokens: 8192, functionCall: true },
+  "qwen2.5-coder-32b-instruct": { category: "code", contextLength: 32768, maxTokens: 8192, functionCall: true },
+  "qwen2.5-coder-7b-instruct": { category: "code", contextLength: 32768, maxTokens: 8192, functionCall: true }
+};
+var VOLCENGINE_DOUBAO_MODELS = {
+  "ark-code-latest": { category: "code", contextLength: 256e3, maxTokens: 32768, functionCall: true },
+  "doubao-seed-2.0-code": { category: "code", contextLength: 256e3, maxTokens: 128e3, vision: true, functionCall: true },
+  "doubao-seed-2.0-lite": { category: "chat", contextLength: 256e3, maxTokens: 32768, functionCall: true }
+};
+var TENCENT_HUNYUAN_MODELS = {
+  "tc-code-latest": { category: "code", functionCall: true },
+  "hunyuan-2.0-instruct": { category: "chat", functionCall: true },
+  "hunyuan-turbos": { category: "chat", functionCall: true }
+};
+var MINIMAX_MODELS = {
+  "minimax-m2.5": { category: "code", contextLength: 204800, functionCall: true },
+  "minimax-m2.5-highspeed": { category: "code", contextLength: 204800, functionCall: true },
+  "minimax-m2.1": { category: "code", contextLength: 204800, functionCall: true },
+  "minimax-m2.1-highspeed": { category: "code", contextLength: 204800, functionCall: true }
+};
+var XIAOMI_MIMO_CANONICAL = {
+  "mimo-v2.5": { category: "code", contextLength: 1e6, maxTokens: 32768, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } },
+  // mimo-v2.5-pro does not support vision (text + reasoning only).
+  "mimo-v2.5-pro": { category: "code", contextLength: 1e6, maxTokens: 32768, vision: false, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } },
+  "mimo-v2-pro": { category: "code", contextLength: 1e6, maxTokens: 32768, vision: true, functionCall: true, reasoning: true, thinkingLevels: ["none", "low", "medium", "high"], thinkingTokenLimit: { min: 0, max: 32768 } },
+  "mimo-v2-flash": { category: "code", contextLength: 256e3, maxTokens: 32768, vision: true, functionCall: true },
+  "mimo-v2-omni": { category: "chat", contextLength: 256e3, maxTokens: 32768, vision: true, functionCall: true }
+};
+var META_LLAMA_MODELS = {
+  "llama-3.3-70b": { category: "chat", contextLength: 131072, maxTokens: 8192, functionCall: true },
+  "llama-3.3-70b-versatile": { category: "chat", contextLength: 131072, maxTokens: 32768, functionCall: true },
+  "llama-3.3-70b-instruct": { category: "chat", contextLength: 131072, maxTokens: 8192, functionCall: true },
+  "llama-3.3-70b-instruct-turbo": { category: "chat", contextLength: 131072, maxTokens: 8192, functionCall: true },
+  "llama-3.1-8b-instant": { category: "chat", contextLength: 131072, maxTokens: 8192, functionCall: true }
+};
+var MISTRAL_MODELS = {
+  "mistral-large-latest": { category: "chat", contextLength: 131072, maxTokens: 8192, functionCall: true },
+  "codestral-latest": { category: "code", contextLength: 256e3, maxTokens: 8192, functionCall: true },
+  "mistral-medium-latest": { category: "chat", contextLength: 131072, maxTokens: 8192, functionCall: true },
+  "mixtral-8x7b-32768": { category: "chat", contextLength: 32768, maxTokens: 32768, functionCall: true }
+};
+var PERPLEXITY_MODELS = {
+  "sonar": { category: "chat", contextLength: 127072, maxTokens: 8192, webSearch: true },
+  "sonar-pro": { category: "chat", contextLength: 2e5, maxTokens: 8192, webSearch: true },
+  "sonar-reasoning-pro": { category: "reasoning", contextLength: 127072, maxTokens: 8192, reasoning: true, webSearch: true }
+};
+var BAIDU_ERNIE_MODELS = {
+  "qianfan-code-latest": { category: "code", contextLength: 98304, maxTokens: 65536, functionCall: true },
+  "ernie-4.5": { category: "chat", functionCall: true },
+  "ernie-3.5": { category: "chat" }
+};
+var KUAISHOU_KAT_MODELS = {
+  "kat-coder-pro-v1": { category: "code", contextLength: 256e3, maxTokens: 128e3, functionCall: true },
+  "kat-coder-air-v1": { category: "code", contextLength: 128e3, maxTokens: 32768, functionCall: true }
+};
+var VENDOR_GROUPS = [
+  { name: "OPENAI_MODELS", entries: OPENAI_MODELS },
+  { name: "ANTHROPIC_MODELS", entries: ANTHROPIC_MODELS },
+  { name: "GEMINI_MODELS", entries: GEMINI_MODELS },
+  { name: "GROK_MODELS", entries: GROK_MODELS },
+  { name: "DEEPSEEK_MODELS", entries: DEEPSEEK_MODELS },
+  { name: "ZHIPU_MODELS", entries: ZHIPU_MODELS },
+  { name: "KIMI_MODELS", entries: KIMI_MODELS },
+  { name: "DASHSCOPE_QWEN_MODELS", entries: DASHSCOPE_QWEN_MODELS },
+  { name: "VOLCENGINE_DOUBAO_MODELS", entries: VOLCENGINE_DOUBAO_MODELS },
+  { name: "TENCENT_HUNYUAN_MODELS", entries: TENCENT_HUNYUAN_MODELS },
+  { name: "MINIMAX_MODELS", entries: MINIMAX_MODELS },
+  { name: "XIAOMI_MIMO_CANONICAL", entries: XIAOMI_MIMO_CANONICAL },
+  { name: "META_LLAMA_MODELS", entries: META_LLAMA_MODELS },
+  { name: "MISTRAL_MODELS", entries: MISTRAL_MODELS },
+  { name: "PERPLEXITY_MODELS", entries: PERPLEXITY_MODELS },
+  { name: "BAIDU_ERNIE_MODELS", entries: BAIDU_ERNIE_MODELS },
+  { name: "KUAISHOU_KAT_MODELS", entries: KUAISHOU_KAT_MODELS }
+];
+function assertNoDuplicateCanonicalIds() {
+  const seen = /* @__PURE__ */ new Map();
+  for (const { name, entries } of VENDOR_GROUPS) {
+    for (const id of Object.keys(entries)) {
+      const prior = seen.get(id);
+      if (prior) {
+        throw new Error(
+          `[canonical-models] duplicate id '${id}' registered in both '${prior}' and '${name}'. Each model must live in exactly one vendor group.`
+        );
+      }
+      seen.set(id, name);
+    }
+  }
+}
+assertNoDuplicateCanonicalIds();
+var KNOWN_MODELS = Object.freeze(
+  Object.assign(
+    /* @__PURE__ */ Object.create(null),
+    ...VENDOR_GROUPS.map((g) => g.entries)
+  )
+);
+function normalizeModelId(rawId) {
+  if (!rawId) return "";
+  const slashIdx = rawId.lastIndexOf("/");
+  const base = slashIdx >= 0 ? rawId.slice(slashIdx + 1) : rawId;
+  const colonIdx = base.indexOf(":");
+  const noSuffix = colonIdx >= 0 ? base.slice(0, colonIdx) : base;
+  return noSuffix.toLowerCase();
+}
+var MODEL_ALIASES = {
+  "gpt-5.6": "gpt-5.6-sol",
+  "deepseek-chat": "deepseek-v3",
+  "deepseek-reasoner": "deepseek-r1"
+};
+function applyAlias(normalizedId) {
+  return MODEL_ALIASES[normalizedId] ?? normalizedId;
+}
+function assertAliasesPointToKnownModels() {
+  for (const [from, to] of Object.entries(MODEL_ALIASES)) {
+    if (!(to in KNOWN_MODELS)) {
+      throw new Error(
+        `[canonical-models] alias '${from}' \u2192 '${to}' points at unknown id. Either add '${to}' to a vendor group or fix the alias.`
+      );
+    }
+  }
+}
+assertAliasesPointToKnownModels();
+function lookupCanonicalCapabilities(rawId, extendedContext = false) {
+  const normalized = normalizeModelId(rawId);
+  if (!normalized) return void 0;
+  const canonicalId = applyAlias(normalized);
+  if (extendedContext) {
+    const extendedKey = `${canonicalId}[1m]`;
+    const extended = KNOWN_MODELS[extendedKey];
+    if (extended) return extended;
+  }
+  return KNOWN_MODELS[canonicalId];
+}
+
 // src/thinking-config.ts
 var EFFORT_RATIO = {
   none: 0.01,
@@ -9,7 +211,7 @@ var EFFORT_RATIO = {
   max: 0.95
 };
 var DEFAULT_MAX_TOKENS = 4096;
-var THINKING_TOKEN_MAP = {
+var LEGACY_THINKING_TOKEN_LIMITS = {
   // Gemini
   "gemini-2\\.5-flash-lite": { min: 512, max: 24576 },
   "gemini-.*-flash": { min: 0, max: 24576 },
@@ -65,6 +267,7 @@ var THINKING_TOKEN_MAP = {
   // Doubao
   "doubao.*think": { min: 0, max: 16384 }
 };
+var THINKING_TOKEN_MAP = LEGACY_THINKING_TOKEN_LIMITS;
 var REASONING_MODEL_PATTERNS = [
   /^o[134](-mini|-preview|-pro)?$/i,
   /^gpt-5(\.\d)?(-pro|-codex|-codex-max)?/i,
@@ -97,8 +300,14 @@ 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;
@@ -191,6 +400,7 @@ export {
   CANNOT_DISABLE_THINKING_PATTERNS,
   DEFAULT_MAX_TOKENS,
   EFFORT_RATIO,
+  LEGACY_THINKING_TOKEN_LIMITS,
   REASONING_MODEL_PATTERNS,
   THINKING_TOKEN_MAP,
   buildAnthropicThinking,
@@ -198,6 +408,7 @@ export {
   buildQwenThinkingConfig,
   calculateThinkingBudget,
   canDisableThinking,
+  findLegacyTokenLimit,
   findTokenLimit,
   getClaudeMaxTokens,
   getOpenAIReasoningEffort,

package/dist/usage-stats-types.d.cts

@@ -46,6 +46,12 @@ interface UsageEventRecord {
     rawUsage: string | null;
     /** Host run-correlation id (owning agent-run, when the host has one). Additive / nullable. */
     runId?: string | null;
+    /** Token-free process-local Route Lease id; never the route bearer token. */
+    routeLeaseId?: string | null;
+    /** Bounded logical Route Lease consumer. */
+    routeLeaseConsumer?: string | null;
+    /** Bounded Route Lease stage attribution. */
+    routeLeaseStageId?: string | null;
     /** Host event-correlation id for this LLM call. NULL when unknown. Additive / nullable. */
     eventId?: string | null;
     /** Safe prompt-cache-key provenance; absent on historical/non-attributed rows. */

package/dist/usage-stats-types.d.ts

@@ -46,6 +46,12 @@ interface UsageEventRecord {
     rawUsage: string | null;
     /** Host run-correlation id (owning agent-run, when the host has one). Additive / nullable. */
     runId?: string | null;
+    /** Token-free process-local Route Lease id; never the route bearer token. */
+    routeLeaseId?: string | null;
+    /** Bounded logical Route Lease consumer. */
+    routeLeaseConsumer?: string | null;
+    /** Bounded Route Lease stage attribution. */
+    routeLeaseStageId?: string | null;
     /** Host event-correlation id for this LLM call. NULL when unknown. Additive / nullable. */
     eventId?: string | null;
     /** Safe prompt-cache-key provenance; absent on historical/non-attributed rows. */

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@omnicross/contracts",
-  "version": "0.1.7",
+  "version": "0.1.9",
   "description": "Dependency-light, host-agnostic contract types + runtime-value helpers shared by the @omnicross/* packages.",
   "license": "MIT",
   "author": "Sayo (https://github.com/Dumoedss)",