@blockrun/clawrouter 0.9.32 → 0.9.33

package/dist/cli.js

@@ -1255,44 +1255,27 @@ var DEFAULT_ROUTING_CONFIG = {
       ]
     }
   },
-  // Eco tier configs - ultra cost-optimized (blockrun/eco)
+  // Eco tier configs - absolute cheapest (blockrun/eco)
   ecoTiers: {
     SIMPLE: {
-      primary: "minimax/minimax-m2.5",
-      // $0.30/$1.20 - cheapest with reasoning
-      fallback: [
-        "moonshot/kimi-k2.5",
-        "nvidia/gpt-oss-120b",
-        "deepseek/deepseek-chat",
-        "google/gemini-2.5-flash"
-      ]
+      primary: "nvidia/gpt-oss-120b",
+      // FREE! $0.00/$0.00
+      fallback: ["google/gemini-2.5-flash", "deepseek/deepseek-chat", "minimax/minimax-m2.5"]
     },
     MEDIUM: {
-      primary: "deepseek/deepseek-chat",
-      // $0.14/$0.28
-      fallback: [
-        "minimax/minimax-m2.5",
-        // $0.30/$1.20 - cheap with reasoning
-        "xai/grok-code-fast-1",
-        "google/gemini-2.5-flash",
-        "moonshot/kimi-k2.5"
-      ]
+      primary: "google/gemini-2.5-flash",
+      // $0.15/$0.60 - cheapest capable
+      fallback: ["deepseek/deepseek-chat", "nvidia/gpt-oss-120b", "minimax/minimax-m2.5"]
     },
     COMPLEX: {
-      primary: "xai/grok-4-0709",
-      // $0.20/$1.50
-      fallback: [
-        "minimax/minimax-m2.5",
-        // $0.30/$1.20 - cheap with reasoning
-        "deepseek/deepseek-chat",
-        "google/gemini-2.5-flash",
-        "openai/gpt-4o-mini"
-      ]
+      primary: "google/gemini-2.5-flash",
+      // $0.15/$0.60 - 1M context handles complexity
+      fallback: ["deepseek/deepseek-chat", "xai/grok-4-0709", "minimax/minimax-m2.5"]
     },
     REASONING: {
-      primary: "minimax/minimax-m2.5",
-      // $0.30/$1.20 - cheapest reasoning model
-      fallback: ["deepseek/deepseek-reasoner", "xai/grok-4-1-fast-reasoning"]
+      primary: "xai/grok-4-1-fast-reasoning",
+      // $0.20/$0.50 - was MORE expensive than AUTO!
+      fallback: ["deepseek/deepseek-reasoner", "minimax/minimax-m2.5"]
     }
   },
   // Premium tier configs - best quality (blockrun/premium)
@@ -1469,6 +1452,8 @@ var MODEL_ALIASES = {
   // Claude - short names (backend uses bare model names without anthropic/ prefix)
   claude: "claude-sonnet-4",
   sonnet: "claude-sonnet-4",
+  "sonnet-4.6": "claude-sonnet-4",
+  "sonnet-4-6": "claude-sonnet-4",
   opus: "claude-opus-4",
   "opus-4": "claude-opus-4",
   haiku: "claude-haiku-4.5",
@@ -4491,10 +4476,12 @@ async function proxyRequest(req, res, apiBase, payFetch, options, routerOpts, de
       console.log(
         `[ClawRouter] Received model: "${parsed.model}" -> normalized: "${normalizedModel}"${wasAlias ? ` -> alias: "${resolvedModel}"` : ""}${routingProfile ? `, profile: ${routingProfile}` : ""}`
       );
-      if (wasAlias && !isRoutingProfile) {
-        parsed.model = resolvedModel;
+      if (!isRoutingProfile) {
+        if (parsed.model !== resolvedModel) {
+          parsed.model = resolvedModel;
+          bodyModified = true;
+        }
         modelId = resolvedModel;
-        bodyModified = true;
       }
       if (isRoutingProfile) {
         if (routingProfile === "free") {
@@ -5004,9 +4991,11 @@ async function proxyRequest(req, res, apiBase, payFetch, options, routerOpts, de
           body: responseBody,
           status: upstream.status,
           headers: responseHeaders,
-          model: modelId
+          model: actualModelUsed
         });
-        console.log(`[ClawRouter] Cached response for ${modelId} (${responseBody.length} bytes)`);
+        console.log(
+          `[ClawRouter] Cached response for ${actualModelUsed} (${responseBody.length} bytes)`
+        );
       }
       try {
         const rspJson = JSON.parse(responseBody.toString());