@talqing/sdk 0.16.0 → 0.16.1

package/dist/cjs/gen/types.gen.d.ts

@@ -6319,7 +6319,7 @@ export type LLMModelSpec = {
     /**
      * Priority
      *
-     * Run this model in its provider's priority (low-latency) lane, where its catalog entry declares one. Steadies the worst case rather than the average, and costs 1.75x-2.5x per token.
+     * Run this model in its provider's priority (low-latency) lane, where its catalog entry declares one; on OpenRouter, any model (standard where no host sells one). Steadies the worst case rather than the average, and costs more per token.
      */
     priority?: boolean;
     /**
@@ -6356,7 +6356,7 @@ export type LLMModelSpecOverride = {
     /**
      * Priority
      *
-     * Run this model in its provider's priority (low-latency) lane, where its catalog entry declares one. Steadies the worst case rather than the average, and costs 1.75x-2.5x per token.
+     * Run this model in its provider's priority (low-latency) lane, where its catalog entry declares one; on OpenRouter, any model (standard where no host sells one). Steadies the worst case rather than the average, and costs more per token.
      */
     priority?: boolean | null;
     /**
@@ -6463,7 +6463,7 @@ export type LLMSpec = {
     /**
      * Priority
      *
-     * Run this model in its provider's priority (low-latency) lane, where its catalog entry declares one. Steadies the worst case rather than the average, and costs 1.75x-2.5x per token.
+     * Run this model in its provider's priority (low-latency) lane, where its catalog entry declares one; on OpenRouter, any model (standard where no host sells one). Steadies the worst case rather than the average, and costs more per token.
      */
     priority?: boolean;
     /**
@@ -6501,7 +6501,7 @@ export type LLMSpecOverride = {
     /**
      * Priority
      *
-     * Run this model in its provider's priority (low-latency) lane, where its catalog entry declares one. Steadies the worst case rather than the average, and costs 1.75x-2.5x per token.
+     * Run this model in its provider's priority (low-latency) lane, where its catalog entry declares one; on OpenRouter, any model (standard where no host sells one). Steadies the worst case rather than the average, and costs more per token.
      */
     priority?: boolean | null;
     /**
@@ -6696,6 +6696,10 @@ export type ModelHostResponse = {
      * Quantization
      */
     quantization: string | null;
+    /**
+     * Service Tier
+     */
+    service_tier: 'priority' | 'ultrafast' | null;
     /**
      * Throughput P50
      */