@blockrun/llm 3.23.2 → 3.24.1

package/README.md

@@ -444,10 +444,13 @@ picked:
 **Search and routing.** xAI search (`x_search` / `web_search` tools, or
 `search` / `searchParameters` not set to `mode: 'off'`) runs on `xai/*` models
 only. When a routed call asks for search, the candidate chain is narrowed to
-its `xai/*` models in the router's order; when the route has none, `route()` /
-`smartChat()` / `smartChatCompletion()` throw before anything is sent or paid.
-For search, calling an `xai/*` model directly (e.g. `xai/grok-4.3`) is the
-dependable path.
+its `xai/*` models in the router's order. The router seldom ranks xAI, so when
+the route has none, the call falls back to the catalog's search-capable xAI
+models (those `/v1/models` tags `search`): cheapest first on `eco` / `auto`,
+most capable first on `premium`. `routing.reasoning` says when that happened.
+Only a catalog with no `xai/*` chat model at all makes `route()` /
+`smartChat()` / `smartChatCompletion()` throw, before anything is sent or paid.
+Pin an `xai/*` model yourself when you want a specific one.
 
 ### TypeScript Types
 

package/dist/index.cjs

@@ -571,7 +571,7 @@ function withCost(body, response, bookedUsd, scheme) {
 // src/client.ts
 var import_accounts2 = require("viem/accounts");
 
-// node_modules/.pnpm/@blockrun+router-core@https+++codeload.github.com+BlockRunAI+router-core+tar.gz+814fd3b6bc788_xq5fkcckbb2p7dtp6dodey4yri/node_modules/@blockrun/router-core/dist/index.js
+// node_modules/.pnpm/@blockrun+router-core@https+++codeload.github.com+BlockRunAI+router-core+tar.gz+27125a3dfe319_a6zjl5xcggrab43wsmb4mvkqym/node_modules/@blockrun/router-core/dist/index.js
 function scoreTokenCount(estimatedTokens, thresholds) {
   if (estimatedTokens < thresholds.simple) {
     return { name: "tokenCount", score: -1, signal: `short (${estimatedTokens} tokens)` };
@@ -1044,6 +1044,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "anthropic/claude-fable-5.1": {
+    contextWindow: 1e6,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
   "anthropic/claude-haiku-4.5": {
     // override: The public catalog's `categories` omit "vision" for this Anthropic model even though the gateway accepts image input for it (the prior hand-maintained snapshot had it, and Anthropic's model card lists it). Without this the vision filter would silently drop it — reported against the catalog; remove once the categories carry vision.
     contextWindow: 2e5,
@@ -1075,6 +1081,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "anthropic/claude-opus-5.5": {
+    contextWindow: 1e6,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
   "anthropic/claude-sonnet-4.5": {
     contextWindow: 2e5,
     maxOutputTokens: 64e3,
@@ -1094,6 +1106,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "anthropic/claude-sonnet-5.5": {
+    contextWindow: 1e6,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
   "cohere/north-mini-code": {
     // supportsTools: not probed — fails closed
     contextWindow: 256e3,
@@ -1113,6 +1131,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: false
   },
+  "deepseek/deepseek-v4-flash-vision-exp": {
+    contextWindow: 1048576,
+    maxOutputTokens: 65536,
+    supportsTools: false,
+    supportsVision: true
+  },
   "deepseek/deepseek-v4-pro": {
     contextWindow: 1048576,
     maxOutputTokens: 65536,
@@ -1173,6 +1197,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "google/gemini-3.8-flash": {
+    contextWindow: 1048576,
+    maxOutputTokens: 65536,
+    supportsTools: true,
+    supportsVision: true
+  },
   "minimax/minimax-m2.7": {
     contextWindow: 204800,
     maxOutputTokens: 16384,
@@ -1198,7 +1228,7 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: false,
     supportsVision: true
   },
-  "nvidia/nemotron-3-nano-30b": {
+  "nvidia/muse-glimmer-30b": {
     // supportsTools: not probed — fails closed
     contextWindow: 131072,
     maxOutputTokens: 16384,
@@ -1206,10 +1236,11 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsVision: false
   },
   "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning": {
+    // override: The catalog tags this model "vision", but a correctly sized probe does not hold up (ClawRouter, 2026-08-31): a 64x64 solid-red PNG was named correctly 1 of 4 times on Base, and on Solana the image was silently dropped and a text model answered "white". An HTTP 200 with a confident wrong answer gives the caller nothing to branch on, so image turns must not be routed here. It is ecoTiers.SIMPLE.fallback[0] since nemotron-3-nano-30b was delisted (2026-09-08); remove once a probe of this size comes back right on both chains.
     contextWindow: 256e3,
     maxOutputTokens: 16384,
     supportsTools: false,
-    supportsVision: true
+    supportsVision: false
   },
   "nvidia/nemotron-3-ultra-550b": {
     // supportsTools: not probed — fails closed
@@ -1267,6 +1298,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: false
   },
+  "openai/gpt-5.1": {
+    contextWindow: 4e5,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
   "openai/gpt-5.2": {
     contextWindow: 4e5,
     maxOutputTokens: 128e3,
@@ -1361,6 +1398,24 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "openai/gpt-6-astra": {
+    contextWindow: 105e4,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
+  "openai/gpt-6-luna": {
+    contextWindow: 105e4,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
+  "openai/gpt-6-sol": {
+    contextWindow: 105e4,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
   "openai/o1": {
     contextWindow: 2e5,
     maxOutputTokens: 1e5,
@@ -1385,6 +1440,13 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: false
   },
+  "poolside/laguna-s-2.1": {
+    // supportsTools: not probed — fails closed
+    contextWindow: 1048576,
+    maxOutputTokens: 131072,
+    supportsTools: false,
+    supportsVision: false
+  },
   "poolside/laguna-xs-2.1": {
     // supportsTools: not probed — fails closed
     contextWindow: 131072,
@@ -1410,10 +1472,24 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: false
   },
-  "tencent/hy3": {
-    contextWindow: 262144,
-    maxOutputTokens: 128e3,
-    supportsTools: true,
+  "qwen/qwen3.8-flash": {
+    contextWindow: 1e6,
+    maxOutputTokens: 131072,
+    supportsTools: false,
+    supportsVision: true
+  },
+  "qwen/qwen3.8-max": {
+    // supportsTools: not probed — fails closed
+    contextWindow: 1e6,
+    maxOutputTokens: 131072,
+    supportsTools: false,
+    supportsVision: false
+  },
+  "tencent/hy4-preview": {
+    // supportsTools: not probed — fails closed
+    contextWindow: 1048576,
+    maxOutputTokens: 64e3,
+    supportsTools: false,
     supportsVision: false
   },
   "xai/grok-4.3": {
@@ -1428,12 +1504,30 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "xai/grok-4.6": {
+    contextWindow: 5e5,
+    maxOutputTokens: 16384,
+    supportsTools: true,
+    supportsVision: true
+  },
+  "xai/grok-4.7": {
+    contextWindow: 5e5,
+    maxOutputTokens: 16384,
+    supportsTools: true,
+    supportsVision: true
+  },
   "xai/grok-build-0.1": {
     contextWindow: 256e3,
     maxOutputTokens: 16384,
     supportsTools: true,
     supportsVision: false
   },
+  "xiaomi/mimo-v2.5": {
+    contextWindow: 1048576,
+    maxOutputTokens: 131072,
+    supportsTools: true,
+    supportsVision: true
+  },
   "xiaomi/mimo-v2.5-pro": {
     contextWindow: 1048576,
     maxOutputTokens: 131072,
@@ -2676,6 +2770,19 @@ function evidenceCandidates(task) {
   }
   return [];
 }
+var VERIFIED_TOOL_FALLBACKS = [
+  "anthropic/claude-fable-5.1",
+  "anthropic/claude-opus-5.5",
+  "anthropic/claude-sonnet-5.5",
+  "google/gemini-3.8-flash",
+  "openai/gpt-5.1",
+  "openai/gpt-6-astra",
+  "openai/gpt-6-luna",
+  "openai/gpt-6-sol",
+  "xai/grok-4.6",
+  "xai/grok-4.7",
+  "xiaomi/mimo-v2.5"
+];
 function isEligible(modelId, features, maxOutputTokens, options) {
   const model = options.modelCapabilities?.[modelId] ?? DEFAULT_MODEL_CAPABILITIES[modelId];
   if (!model) return true;
@@ -2722,8 +2829,17 @@ var PortfolioStrategy = class {
     const tierConfig = tierConfigs[targetTier];
     const configuredCandidates = tierConfig ? getFallbackChain(targetTier, tierConfigs) : [];
     const unavailable = new Set(options.unavailableModels ?? []);
+    const verifiedToolFallbacks = features.needsTools ? VERIFIED_TOOL_FALLBACKS.filter((model2) => options.modelPricing.has(model2)) : [];
+    if (options.routingProfile !== "premium") {
+      const fallbackCost = (model2) => estimatedCost(model2, options, features.estimatedInputTokens, maxOutputTokens);
+      verifiedToolFallbacks.sort((a, b) => fallbackCost(a) - fallbackCost(b));
+    }
     const chain = [
-      .../* @__PURE__ */ new Set([...configuredCandidates, ...evidenceCandidates(features.taskType)])
+      .../* @__PURE__ */ new Set([
+        ...configuredCandidates,
+        ...evidenceCandidates(features.taskType),
+        ...verifiedToolFallbacks
+      ])
     ].filter(
       (model2) => typeof model2 === "string" && model2.length > 0 && !unavailable.has(model2)
     );
@@ -4026,8 +4142,13 @@ var DEFAULT_ROUTING_CONFIG = {
       primary: "nvidia/nemotron-3.5-lightning",
       // FREE — NVIDIA free tier flagship, 1M ctx
       fallback: [
-        "nvidia/nemotron-3-nano-30b",
-        // FREE — fastest free model (~121 tok/s)
+        // Was nvidia/nemotron-3-nano-30b until blockrun delisted it on 2026-09-08
+        // (NVIDIA deprovisioned it for the account — a structured per-account
+        // 404). nano-omni is blockrun's own redirect target for that id, so this
+        // follows the same rule as the head: the router and the gateway name the
+        // same model.
+        "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning",
+        // FREE — nano-30b's successor, 256K ctx
         // The free head keeps rotting with NVIDIA's hosting (deepseek-v4-flash
         // 410 2026-08-12, seed-oss-36b 410 2026-08-03, gpt-oss-120b/20b 400
         // 2026-08-21, and on 2026-08-30 FOUR of the five visible free models at
@@ -4359,6 +4480,24 @@ function requestsSearch(options) {
   if (options.searchParameters !== void 0) return options.searchParameters.mode !== "off";
   return options.search === true;
 }
+var searchModelsByCatalog = /* @__PURE__ */ new WeakMap();
+function registerSearchModels(pricing, models) {
+  const xaiChat = models.filter(
+    (m) => m.id.startsWith("xai/") && m.available !== false && Array.isArray(m.categories) && m.categories.includes("chat") && pricing.has(m.id)
+  );
+  const tagged = xaiChat.filter((m) => m.categories.includes("search"));
+  searchModelsByCatalog.set(pricing, (tagged.length > 0 ? tagged : xaiChat).map((m) => m.id));
+}
+function searchFallbackChain(modelPricing, routingProfile) {
+  const pool = searchModelsByCatalog.get(modelPricing) ?? [...modelPricing.keys()].filter((id) => id.startsWith("xai/"));
+  const price = (id) => {
+    const p = modelPricing.get(id);
+    return p ? p.inputPrice + p.outputPrice : Number.POSITIVE_INFINITY;
+  };
+  const sorted = [...pool].sort((a, b) => price(a) - price(b) || b.localeCompare(a, void 0, { numeric: true }));
+  if (routingProfile !== "premium") return sorted;
+  return [...pool].sort((a, b) => price(b) - price(a) || b.localeCompare(a, void 0, { numeric: true }));
+}
 function routingProfileForModel(model) {
   return AUTO_ROUTING_PROFILES[model.toLowerCase()];
 }
@@ -4406,11 +4545,16 @@ function routeWithCatalog(prompt, systemPrompt, maxOutputTokens, modelPricing, o
     const resolved = !id.startsWith("free/") ? id : modelPricing.has(`nvidia/${id.slice(5)}`) ? `nvidia/${id.slice(5)}` : null;
     if (resolved && !callable.includes(resolved)) callable.push(resolved);
   }
+  let searchFallback = false;
   if (requestsSearch(options)) {
-    const searchable = callable.filter((id) => id.startsWith("xai/"));
+    let searchable = callable.filter((id) => id.startsWith("xai/"));
+    if (searchable.length === 0) {
+      searchable = searchFallbackChain(modelPricing, options.routingProfile);
+      searchFallback = true;
+    }
     if (searchable.length === 0) {
       throw new Error(
-        `Search (x_search / web_search tools or searchParameters) runs on xai/* models only, and the ${options.routingProfile ?? "auto"} route has no xai/* candidate for this request. Call an xai/* model directly (e.g. xai/grok-4.3) or drop the search options.`
+        `Search (x_search / web_search tools or searchParameters) runs on xai/* models only, and the catalog lists no available xai/* chat model. Call an xai/* model directly or drop the search options.`
       );
     }
     callable.splice(0, callable.length, ...searchable);
@@ -4447,7 +4591,7 @@ function routeWithCatalog(prompt, systemPrompt, maxOutputTokens, modelPricing, o
     costEstimate,
     savings,
     model,
-    reasoning: model === decision.model ? decision.reasoning : `${decision.reasoning} | catalog fallback: ${model}`,
+    reasoning: searchFallback ? `${decision.reasoning} | search runs on xai/* only; the route had none, so the ${options.routingProfile === "premium" ? "most capable" : "cheapest"} search-capable xAI model: ${model}` : model === decision.model ? decision.reasoning : `${decision.reasoning} | catalog fallback: ${model}`,
     candidates: availableCandidates,
     candidateScores: decision.candidateScores?.filter(
       (score) => availableCandidates.includes(score.model)
@@ -4833,7 +4977,7 @@ function getCostSummary() {
 }
 
 // src/version.ts
-var SDK_VERSION = "3.23.2";
+var SDK_VERSION = "3.24.1";
 var USER_AGENT = `blockrun-ts/${SDK_VERSION}`;
 
 // src/client.ts
@@ -5104,6 +5248,7 @@ var LLMClient = class _LLMClient {
         pricing.set(model.id, { inputPrice, outputPrice });
       }
     }
+    registerSearchModels(pricing, models);
     return pricing;
   }
   /**
@@ -12567,6 +12712,7 @@ var SolanaLLMClient = class {
           ...(model.billingMode ?? raw.billing_mode) === "flat" && flatPrice > 0 ? { flatPrice } : {}
         });
       }
+      registerSearchModels(pricing, models);
       return pricing;
     })();
     try {

package/dist/index.d.cts

@@ -2953,8 +2953,9 @@ interface SolanaBatchOptions {
      * re-read, in addition to the one-line log the SDK writes to stderr for each.
      * A callback that throws, or returns a promise that rejects, is logged and
      * ignored; it never affects the payment, and its promise is not awaited.
+     * Any other return value is ignored, so `(e) => events.push(e)` is fine.
      */
-    onEvent?: (event: SolanaBatchEvent) => void | PromiseLike<void>;
+    onEvent?: (event: SolanaBatchEvent) => unknown;
 }
 /** Default {@link SolanaBatchOptions.rateLimit}. */
 declare const DEFAULT_BATCH_RATE_LIMIT: {

package/dist/index.d.ts

@@ -2953,8 +2953,9 @@ interface SolanaBatchOptions {
      * re-read, in addition to the one-line log the SDK writes to stderr for each.
      * A callback that throws, or returns a promise that rejects, is logged and
      * ignored; it never affects the payment, and its promise is not awaited.
+     * Any other return value is ignored, so `(e) => events.push(e)` is fine.
      */
-    onEvent?: (event: SolanaBatchEvent) => void | PromiseLike<void>;
+    onEvent?: (event: SolanaBatchEvent) => unknown;
 }
 /** Default {@link SolanaBatchOptions.rateLimit}. */
 declare const DEFAULT_BATCH_RATE_LIMIT: {

package/dist/index.js

@@ -440,7 +440,7 @@ function withCost(body, response, bookedUsd, scheme) {
 // src/client.ts
 import { privateKeyToAccount } from "viem/accounts";
 
-// node_modules/.pnpm/@blockrun+router-core@https+++codeload.github.com+BlockRunAI+router-core+tar.gz+814fd3b6bc788_xq5fkcckbb2p7dtp6dodey4yri/node_modules/@blockrun/router-core/dist/index.js
+// node_modules/.pnpm/@blockrun+router-core@https+++codeload.github.com+BlockRunAI+router-core+tar.gz+27125a3dfe319_a6zjl5xcggrab43wsmb4mvkqym/node_modules/@blockrun/router-core/dist/index.js
 function scoreTokenCount(estimatedTokens, thresholds) {
   if (estimatedTokens < thresholds.simple) {
     return { name: "tokenCount", score: -1, signal: `short (${estimatedTokens} tokens)` };
@@ -913,6 +913,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "anthropic/claude-fable-5.1": {
+    contextWindow: 1e6,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
   "anthropic/claude-haiku-4.5": {
     // override: The public catalog's `categories` omit "vision" for this Anthropic model even though the gateway accepts image input for it (the prior hand-maintained snapshot had it, and Anthropic's model card lists it). Without this the vision filter would silently drop it — reported against the catalog; remove once the categories carry vision.
     contextWindow: 2e5,
@@ -944,6 +950,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "anthropic/claude-opus-5.5": {
+    contextWindow: 1e6,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
   "anthropic/claude-sonnet-4.5": {
     contextWindow: 2e5,
     maxOutputTokens: 64e3,
@@ -963,6 +975,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "anthropic/claude-sonnet-5.5": {
+    contextWindow: 1e6,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
   "cohere/north-mini-code": {
     // supportsTools: not probed — fails closed
     contextWindow: 256e3,
@@ -982,6 +1000,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: false
   },
+  "deepseek/deepseek-v4-flash-vision-exp": {
+    contextWindow: 1048576,
+    maxOutputTokens: 65536,
+    supportsTools: false,
+    supportsVision: true
+  },
   "deepseek/deepseek-v4-pro": {
     contextWindow: 1048576,
     maxOutputTokens: 65536,
@@ -1042,6 +1066,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "google/gemini-3.8-flash": {
+    contextWindow: 1048576,
+    maxOutputTokens: 65536,
+    supportsTools: true,
+    supportsVision: true
+  },
   "minimax/minimax-m2.7": {
     contextWindow: 204800,
     maxOutputTokens: 16384,
@@ -1067,7 +1097,7 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: false,
     supportsVision: true
   },
-  "nvidia/nemotron-3-nano-30b": {
+  "nvidia/muse-glimmer-30b": {
     // supportsTools: not probed — fails closed
     contextWindow: 131072,
     maxOutputTokens: 16384,
@@ -1075,10 +1105,11 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsVision: false
   },
   "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning": {
+    // override: The catalog tags this model "vision", but a correctly sized probe does not hold up (ClawRouter, 2026-08-31): a 64x64 solid-red PNG was named correctly 1 of 4 times on Base, and on Solana the image was silently dropped and a text model answered "white". An HTTP 200 with a confident wrong answer gives the caller nothing to branch on, so image turns must not be routed here. It is ecoTiers.SIMPLE.fallback[0] since nemotron-3-nano-30b was delisted (2026-09-08); remove once a probe of this size comes back right on both chains.
     contextWindow: 256e3,
     maxOutputTokens: 16384,
     supportsTools: false,
-    supportsVision: true
+    supportsVision: false
   },
   "nvidia/nemotron-3-ultra-550b": {
     // supportsTools: not probed — fails closed
@@ -1136,6 +1167,12 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: false
   },
+  "openai/gpt-5.1": {
+    contextWindow: 4e5,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
   "openai/gpt-5.2": {
     contextWindow: 4e5,
     maxOutputTokens: 128e3,
@@ -1230,6 +1267,24 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "openai/gpt-6-astra": {
+    contextWindow: 105e4,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
+  "openai/gpt-6-luna": {
+    contextWindow: 105e4,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
+  "openai/gpt-6-sol": {
+    contextWindow: 105e4,
+    maxOutputTokens: 128e3,
+    supportsTools: true,
+    supportsVision: true
+  },
   "openai/o1": {
     contextWindow: 2e5,
     maxOutputTokens: 1e5,
@@ -1254,6 +1309,13 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: false
   },
+  "poolside/laguna-s-2.1": {
+    // supportsTools: not probed — fails closed
+    contextWindow: 1048576,
+    maxOutputTokens: 131072,
+    supportsTools: false,
+    supportsVision: false
+  },
   "poolside/laguna-xs-2.1": {
     // supportsTools: not probed — fails closed
     contextWindow: 131072,
@@ -1279,10 +1341,24 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: false
   },
-  "tencent/hy3": {
-    contextWindow: 262144,
-    maxOutputTokens: 128e3,
-    supportsTools: true,
+  "qwen/qwen3.8-flash": {
+    contextWindow: 1e6,
+    maxOutputTokens: 131072,
+    supportsTools: false,
+    supportsVision: true
+  },
+  "qwen/qwen3.8-max": {
+    // supportsTools: not probed — fails closed
+    contextWindow: 1e6,
+    maxOutputTokens: 131072,
+    supportsTools: false,
+    supportsVision: false
+  },
+  "tencent/hy4-preview": {
+    // supportsTools: not probed — fails closed
+    contextWindow: 1048576,
+    maxOutputTokens: 64e3,
+    supportsTools: false,
     supportsVision: false
   },
   "xai/grok-4.3": {
@@ -1297,12 +1373,30 @@ var DEFAULT_MODEL_CAPABILITIES = Object.freeze({
     supportsTools: true,
     supportsVision: true
   },
+  "xai/grok-4.6": {
+    contextWindow: 5e5,
+    maxOutputTokens: 16384,
+    supportsTools: true,
+    supportsVision: true
+  },
+  "xai/grok-4.7": {
+    contextWindow: 5e5,
+    maxOutputTokens: 16384,
+    supportsTools: true,
+    supportsVision: true
+  },
   "xai/grok-build-0.1": {
     contextWindow: 256e3,
     maxOutputTokens: 16384,
     supportsTools: true,
     supportsVision: false
   },
+  "xiaomi/mimo-v2.5": {
+    contextWindow: 1048576,
+    maxOutputTokens: 131072,
+    supportsTools: true,
+    supportsVision: true
+  },
   "xiaomi/mimo-v2.5-pro": {
     contextWindow: 1048576,
     maxOutputTokens: 131072,
@@ -2545,6 +2639,19 @@ function evidenceCandidates(task) {
   }
   return [];
 }
+var VERIFIED_TOOL_FALLBACKS = [
+  "anthropic/claude-fable-5.1",
+  "anthropic/claude-opus-5.5",
+  "anthropic/claude-sonnet-5.5",
+  "google/gemini-3.8-flash",
+  "openai/gpt-5.1",
+  "openai/gpt-6-astra",
+  "openai/gpt-6-luna",
+  "openai/gpt-6-sol",
+  "xai/grok-4.6",
+  "xai/grok-4.7",
+  "xiaomi/mimo-v2.5"
+];
 function isEligible(modelId, features, maxOutputTokens, options) {
   const model = options.modelCapabilities?.[modelId] ?? DEFAULT_MODEL_CAPABILITIES[modelId];
   if (!model) return true;
@@ -2591,8 +2698,17 @@ var PortfolioStrategy = class {
     const tierConfig = tierConfigs[targetTier];
     const configuredCandidates = tierConfig ? getFallbackChain(targetTier, tierConfigs) : [];
     const unavailable = new Set(options.unavailableModels ?? []);
+    const verifiedToolFallbacks = features.needsTools ? VERIFIED_TOOL_FALLBACKS.filter((model2) => options.modelPricing.has(model2)) : [];
+    if (options.routingProfile !== "premium") {
+      const fallbackCost = (model2) => estimatedCost(model2, options, features.estimatedInputTokens, maxOutputTokens);
+      verifiedToolFallbacks.sort((a, b) => fallbackCost(a) - fallbackCost(b));
+    }
     const chain = [
-      .../* @__PURE__ */ new Set([...configuredCandidates, ...evidenceCandidates(features.taskType)])
+      .../* @__PURE__ */ new Set([
+        ...configuredCandidates,
+        ...evidenceCandidates(features.taskType),
+        ...verifiedToolFallbacks
+      ])
     ].filter(
       (model2) => typeof model2 === "string" && model2.length > 0 && !unavailable.has(model2)
     );
@@ -3895,8 +4011,13 @@ var DEFAULT_ROUTING_CONFIG = {
       primary: "nvidia/nemotron-3.5-lightning",
       // FREE — NVIDIA free tier flagship, 1M ctx
       fallback: [
-        "nvidia/nemotron-3-nano-30b",
-        // FREE — fastest free model (~121 tok/s)
+        // Was nvidia/nemotron-3-nano-30b until blockrun delisted it on 2026-09-08
+        // (NVIDIA deprovisioned it for the account — a structured per-account
+        // 404). nano-omni is blockrun's own redirect target for that id, so this
+        // follows the same rule as the head: the router and the gateway name the
+        // same model.
+        "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning",
+        // FREE — nano-30b's successor, 256K ctx
         // The free head keeps rotting with NVIDIA's hosting (deepseek-v4-flash
         // 410 2026-08-12, seed-oss-36b 410 2026-08-03, gpt-oss-120b/20b 400
         // 2026-08-21, and on 2026-08-30 FOUR of the five visible free models at
@@ -4228,6 +4349,24 @@ function requestsSearch(options) {
   if (options.searchParameters !== void 0) return options.searchParameters.mode !== "off";
   return options.search === true;
 }
+var searchModelsByCatalog = /* @__PURE__ */ new WeakMap();
+function registerSearchModels(pricing, models) {
+  const xaiChat = models.filter(
+    (m) => m.id.startsWith("xai/") && m.available !== false && Array.isArray(m.categories) && m.categories.includes("chat") && pricing.has(m.id)
+  );
+  const tagged = xaiChat.filter((m) => m.categories.includes("search"));
+  searchModelsByCatalog.set(pricing, (tagged.length > 0 ? tagged : xaiChat).map((m) => m.id));
+}
+function searchFallbackChain(modelPricing, routingProfile) {
+  const pool = searchModelsByCatalog.get(modelPricing) ?? [...modelPricing.keys()].filter((id) => id.startsWith("xai/"));
+  const price = (id) => {
+    const p = modelPricing.get(id);
+    return p ? p.inputPrice + p.outputPrice : Number.POSITIVE_INFINITY;
+  };
+  const sorted = [...pool].sort((a, b) => price(a) - price(b) || b.localeCompare(a, void 0, { numeric: true }));
+  if (routingProfile !== "premium") return sorted;
+  return [...pool].sort((a, b) => price(b) - price(a) || b.localeCompare(a, void 0, { numeric: true }));
+}
 function routingProfileForModel(model) {
   return AUTO_ROUTING_PROFILES[model.toLowerCase()];
 }
@@ -4275,11 +4414,16 @@ function routeWithCatalog(prompt, systemPrompt, maxOutputTokens, modelPricing, o
     const resolved = !id.startsWith("free/") ? id : modelPricing.has(`nvidia/${id.slice(5)}`) ? `nvidia/${id.slice(5)}` : null;
     if (resolved && !callable.includes(resolved)) callable.push(resolved);
   }
+  let searchFallback = false;
   if (requestsSearch(options)) {
-    const searchable = callable.filter((id) => id.startsWith("xai/"));
+    let searchable = callable.filter((id) => id.startsWith("xai/"));
+    if (searchable.length === 0) {
+      searchable = searchFallbackChain(modelPricing, options.routingProfile);
+      searchFallback = true;
+    }
     if (searchable.length === 0) {
       throw new Error(
-        `Search (x_search / web_search tools or searchParameters) runs on xai/* models only, and the ${options.routingProfile ?? "auto"} route has no xai/* candidate for this request. Call an xai/* model directly (e.g. xai/grok-4.3) or drop the search options.`
+        `Search (x_search / web_search tools or searchParameters) runs on xai/* models only, and the catalog lists no available xai/* chat model. Call an xai/* model directly or drop the search options.`
       );
     }
     callable.splice(0, callable.length, ...searchable);
@@ -4316,7 +4460,7 @@ function routeWithCatalog(prompt, systemPrompt, maxOutputTokens, modelPricing, o
     costEstimate,
     savings,
     model,
-    reasoning: model === decision.model ? decision.reasoning : `${decision.reasoning} | catalog fallback: ${model}`,
+    reasoning: searchFallback ? `${decision.reasoning} | search runs on xai/* only; the route had none, so the ${options.routingProfile === "premium" ? "most capable" : "cheapest"} search-capable xAI model: ${model}` : model === decision.model ? decision.reasoning : `${decision.reasoning} | catalog fallback: ${model}`,
     candidates: availableCandidates,
     candidateScores: decision.candidateScores?.filter(
       (score) => availableCandidates.includes(score.model)
@@ -4702,7 +4846,7 @@ function getCostSummary() {
 }
 
 // src/version.ts
-var SDK_VERSION = "3.23.2";
+var SDK_VERSION = "3.24.1";
 var USER_AGENT = `blockrun-ts/${SDK_VERSION}`;
 
 // src/client.ts
@@ -4973,6 +5117,7 @@ var LLMClient = class _LLMClient {
         pricing.set(model.id, { inputPrice, outputPrice });
       }
     }
+    registerSearchModels(pricing, models);
     return pricing;
   }
   /**
@@ -12436,6 +12581,7 @@ var SolanaLLMClient = class {
           ...(model.billingMode ?? raw.billing_mode) === "flat" && flatPrice > 0 ? { flatPrice } : {}
         });
       }
+      registerSearchModels(pricing, models);
       return pricing;
     })();
     try {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@blockrun/llm",
-  "version": "3.23.2",
+  "version": "3.24.1",
   "type": "module",
   "description": "TypeScript SDK for BlockRun - every frontier model behind one OpenAI-compatible client. Pay with a BlockRun API key or per-call in USDC. Smart routing picks the cheapest capable model. No rate limits.",
   "main": "dist/index.cjs",
@@ -58,7 +58,7 @@
   },
   "devDependencies": {
     "@blockrun/core": "^0.1.0",
-    "@blockrun/router-core": "https://codeload.github.com/BlockRunAI/router-core/tar.gz/814fd3b6bc7880a84bbe030bd08c98e8670031e7",
+    "@blockrun/router-core": "https://codeload.github.com/BlockRunAI/router-core/tar.gz/27125a3dfe319a6f422a0c0f41bf1666b0dc61a9",
     "@eslint/js": "^10.0.1",
     "@solana/kit": "^5.5.1",
     "@types/node": "^26.4.1",