smoltalk 0.14.1 → 0.14.2

package/dist/models.d.ts

@@ -176,6 +176,13 @@ export declare const textToSpeechModels: readonly [{
     readonly perCharacterCost: 0.00004;
     readonly maxInputChars: 200;
     readonly formats: readonly ["wav"];
+}, {
+    readonly type: "text-to-speech";
+    readonly modelName: "gemini-3.1-flash-tts-preview";
+    readonly provider: "google";
+    readonly inputTokenCost: 1;
+    readonly outputAudioTokenCost: 20;
+    readonly formats: readonly ["pcm", "wav"];
 }, {
     readonly type: "text-to-speech";
     readonly modelName: "gemini-2.5-flash-preview-tts";
@@ -688,7 +695,7 @@ export declare const textModels: readonly [{
         readonly inputTokenCost: 5;
         readonly cachedInputTokenCost: 0.5;
         readonly outputTokenCost: 22.5;
-        readonly thresholdTokens: 200000;
+        readonly thresholdTokens: 272000;
     };
     readonly reasoning: {
         readonly levels: readonly ["none", "low", "medium", "high", "xhigh"];
@@ -778,7 +785,7 @@ export declare const textModels: readonly [{
     readonly longContext: {
         readonly inputTokenCost: 60;
         readonly outputTokenCost: 270;
-        readonly thresholdTokens: 200000;
+        readonly thresholdTokens: 272000;
     };
     readonly reasoning: {
         readonly levels: readonly ["medium", "high", "xhigh"];
@@ -813,7 +820,7 @@ export declare const textModels: readonly [{
         readonly inputTokenCost: 10;
         readonly cachedInputTokenCost: 1;
         readonly outputTokenCost: 45;
-        readonly thresholdTokens: 200000;
+        readonly thresholdTokens: 272000;
     };
     readonly reasoning: {
         readonly levels: readonly ["none", "low", "medium", "high", "xhigh"];
@@ -845,7 +852,7 @@ export declare const textModels: readonly [{
     readonly longContext: {
         readonly inputTokenCost: 60;
         readonly outputTokenCost: 270;
-        readonly thresholdTokens: 200000;
+        readonly thresholdTokens: 272000;
     };
     readonly reasoning: {
         readonly levels: readonly ["none", "low", "medium", "high", "xhigh"];
@@ -866,6 +873,41 @@ export declare const textModels: readonly [{
     readonly structuredOutput: true;
     readonly temperatureSupported: false;
     readonly provider: "openai-responses";
+}, {
+    readonly type: "text";
+    readonly modelName: "gpt-6-astra";
+    readonly description: "GPT-6 Astra is OpenAI's most capable model, built for the hardest end-to-end work: complex reasoning, coding, computer use, research, and document creation. 1M context window. Standard pricing for ≤272K input tokens; prompts above that are billed at 2x input/cache and 1.5x output for the whole request. Knowledge cutoff: April 2026.";
+    readonly maxInputTokens: 1050000;
+    readonly maxOutputTokens: 128000;
+    readonly inputTokenCost: 10;
+    readonly cachedInputTokenCost: 1;
+    readonly outputTokenCost: 50;
+    readonly outputTokensPerSecond: 69;
+    readonly longContext: {
+        readonly inputTokenCost: 20;
+        readonly cachedInputTokenCost: 2;
+        readonly outputTokenCost: 75;
+        readonly thresholdTokens: 272000;
+    };
+    readonly reasoning: {
+        readonly levels: readonly ["low", "medium", "high", "xhigh", "max"];
+        readonly defaultLevel: "medium";
+        readonly canDisable: false;
+        readonly outputsThinking: false;
+        readonly outputsSignatures: false;
+    };
+    readonly modalities: {
+        readonly input: readonly ["text", "image", "pdf"];
+        readonly output: readonly ["text"];
+    };
+    readonly knowledge: "2026-04-30";
+    readonly releaseDate: "2026-09-04";
+    readonly lastUpdated: "2026-09-04";
+    readonly family: "gpt";
+    readonly openWeights: false;
+    readonly structuredOutput: true;
+    readonly temperatureSupported: false;
+    readonly provider: "openai";
 }, {
     readonly type: "text";
     readonly modelName: "gpt-5.6-sol";
@@ -875,12 +917,12 @@ export declare const textModels: readonly [{
     readonly inputTokenCost: 4;
     readonly cachedInputTokenCost: 0.4;
     readonly outputTokenCost: 20;
-    readonly outputTokensPerSecond: 71;
+    readonly outputTokensPerSecond: 77;
     readonly longContext: {
         readonly inputTokenCost: 8;
         readonly cachedInputTokenCost: 0.8;
         readonly outputTokenCost: 30;
-        readonly thresholdTokens: 200000;
+        readonly thresholdTokens: 272000;
     };
     readonly reasoning: {
         readonly levels: readonly ["none", "low", "medium", "high", "xhigh", "max"];
@@ -910,12 +952,12 @@ export declare const textModels: readonly [{
     readonly inputTokenCost: 2;
     readonly cachedInputTokenCost: 0.2;
     readonly outputTokenCost: 12;
-    readonly outputTokensPerSecond: 98;
+    readonly outputTokensPerSecond: 106;
     readonly longContext: {
         readonly inputTokenCost: 4;
         readonly cachedInputTokenCost: 0.4;
         readonly outputTokenCost: 18;
-        readonly thresholdTokens: 200000;
+        readonly thresholdTokens: 272000;
     };
     readonly reasoning: {
         readonly levels: readonly ["none", "low", "medium", "high", "xhigh", "max"];
@@ -945,12 +987,12 @@ export declare const textModels: readonly [{
     readonly inputTokenCost: 0.2;
     readonly cachedInputTokenCost: 0.02;
     readonly outputTokenCost: 1.2;
-    readonly outputTokensPerSecond: 116;
+    readonly outputTokensPerSecond: 165;
     readonly longContext: {
         readonly inputTokenCost: 0.4;
         readonly cachedInputTokenCost: 0.04;
         readonly outputTokenCost: 1.8;
-        readonly thresholdTokens: 200000;
+        readonly thresholdTokens: 272000;
     };
     readonly reasoning: {
         readonly levels: readonly ["none", "low", "medium", "high", "xhigh", "max"];
@@ -980,7 +1022,7 @@ export declare const textModels: readonly [{
     readonly inputTokenCost: 2;
     readonly cachedInputTokenCost: 0.2;
     readonly outputTokenCost: 12;
-    readonly outputTokensPerSecond: 133;
+    readonly outputTokensPerSecond: 124;
     readonly longContext: {
         readonly inputTokenCost: 4;
         readonly cachedInputTokenCost: 0.4;
@@ -1043,7 +1085,7 @@ export declare const textModels: readonly [{
     readonly inputTokenCost: 0.75;
     readonly cachedInputTokenCost: 0.075;
     readonly outputTokenCost: 3.75;
-    readonly outputTokensPerSecond: 312;
+    readonly outputTokensPerSecond: 329;
     readonly inputAudioTokenCost: 1.5;
     readonly reasoning: {
         readonly levels: readonly ["low", "medium", "high"];
@@ -1455,6 +1497,7 @@ export declare const textModels: readonly [{
     readonly cachedInputTokenCost: 0.25;
     readonly cacheCreationInputTokenCost: 12.5;
     readonly outputTokenCost: 50;
+    readonly outputTokensPerSecond: 69;
     readonly reasoning: {
         readonly thinkingStyle: "adaptive";
         readonly levels: readonly ["low", "medium", "high", "xhigh", "max"];
@@ -1484,6 +1527,7 @@ export declare const textModels: readonly [{
     readonly cachedInputTokenCost: 0.5;
     readonly cacheCreationInputTokenCost: 6.25;
     readonly outputTokenCost: 25;
+    readonly outputTokensPerSecond: 59;
     readonly reasoning: {
         readonly thinkingStyle: "adaptive";
         readonly levels: readonly ["low", "medium", "high", "xhigh", "max"];
@@ -1625,7 +1669,7 @@ export declare const textModels: readonly [{
     readonly cachedInputTokenCost: 0.2;
     readonly cacheCreationInputTokenCost: 2.5;
     readonly outputTokenCost: 10;
-    readonly outputTokensPerSecond: 71;
+    readonly outputTokensPerSecond: 81;
     readonly reasoning: {
         readonly thinkingStyle: "adaptive";
         readonly levels: readonly ["low", "medium", "high", "xhigh", "max"];

package/dist/models.js

@@ -92,6 +92,14 @@ export const textToSpeechModels = [
     },
     // Gemini TTS is token-billed (text input + audio output). No maxInputChars:
     // Gemini documents a 32k-token context, and characters are not a sound proxy.
+    {
+        type: "text-to-speech",
+        modelName: "gemini-3.1-flash-tts-preview",
+        provider: "google",
+        inputTokenCost: 1.0, // $/1M text-input tokens, verified 2026-09-21
+        outputAudioTokenCost: 20.0, // $/1M audio-output tokens
+        formats: ["pcm", "wav"],
+    },
     {
         type: "text-to-speech",
         modelName: "gemini-2.5-flash-preview-tts",
@@ -626,7 +634,7 @@ export const textModels = [
             inputTokenCost: 5,
             cachedInputTokenCost: 0.5,
             outputTokenCost: 22.5,
-            thresholdTokens: 200000,
+            thresholdTokens: 272000,
         },
         reasoning: {
             levels: ["none", "low", "medium", "high", "xhigh"],
@@ -719,7 +727,7 @@ export const textModels = [
         longContext: {
             inputTokenCost: 60,
             outputTokenCost: 270,
-            thresholdTokens: 200000,
+            thresholdTokens: 272000,
         },
         reasoning: {
             levels: ["medium", "high", "xhigh"],
@@ -755,7 +763,7 @@ export const textModels = [
             inputTokenCost: 10,
             cachedInputTokenCost: 1,
             outputTokenCost: 45,
-            thresholdTokens: 200000,
+            thresholdTokens: 272000,
         },
         reasoning: {
             levels: ["none", "low", "medium", "high", "xhigh"],
@@ -788,7 +796,7 @@ export const textModels = [
         longContext: {
             inputTokenCost: 60,
             outputTokenCost: 270,
-            thresholdTokens: 200000,
+            thresholdTokens: 272000,
         },
         reasoning: {
             levels: ["none", "low", "medium", "high", "xhigh"],
@@ -810,6 +818,42 @@ export const textModels = [
         temperatureSupported: false,
         provider: "openai-responses",
     },
+    {
+        type: "text",
+        modelName: "gpt-6-astra",
+        description: "GPT-6 Astra is OpenAI's most capable model, built for the hardest end-to-end work: complex reasoning, coding, computer use, research, and document creation. 1M context window. Standard pricing for ≤272K input tokens; prompts above that are billed at 2x input/cache and 1.5x output for the whole request. Knowledge cutoff: April 2026.",
+        maxInputTokens: 1050000,
+        maxOutputTokens: 128000,
+        inputTokenCost: 10,
+        cachedInputTokenCost: 1,
+        outputTokenCost: 50,
+        outputTokensPerSecond: 69,
+        longContext: {
+            inputTokenCost: 20,
+            cachedInputTokenCost: 2,
+            outputTokenCost: 75,
+            thresholdTokens: 272000,
+        },
+        reasoning: {
+            levels: ["low", "medium", "high", "xhigh", "max"],
+            defaultLevel: "medium",
+            canDisable: false,
+            outputsThinking: false,
+            outputsSignatures: false,
+        },
+        modalities: {
+            input: ["text", "image", "pdf"],
+            output: ["text"],
+        },
+        knowledge: "2026-04-30",
+        releaseDate: "2026-09-04",
+        lastUpdated: "2026-09-04",
+        family: "gpt",
+        openWeights: false,
+        structuredOutput: true,
+        temperatureSupported: false,
+        provider: "openai",
+    },
     {
         type: "text",
         modelName: "gpt-5.6-sol",
@@ -819,12 +863,12 @@ export const textModels = [
         inputTokenCost: 4,
         cachedInputTokenCost: 0.4,
         outputTokenCost: 20,
-        outputTokensPerSecond: 71,
+        outputTokensPerSecond: 77,
         longContext: {
             inputTokenCost: 8,
             cachedInputTokenCost: 0.8,
             outputTokenCost: 30,
-            thresholdTokens: 200000,
+            thresholdTokens: 272000,
         },
         reasoning: {
             levels: ["none", "low", "medium", "high", "xhigh", "max"],
@@ -855,12 +899,12 @@ export const textModels = [
         inputTokenCost: 2,
         cachedInputTokenCost: 0.2,
         outputTokenCost: 12,
-        outputTokensPerSecond: 98,
+        outputTokensPerSecond: 106,
         longContext: {
             inputTokenCost: 4,
             cachedInputTokenCost: 0.4,
             outputTokenCost: 18,
-            thresholdTokens: 200000,
+            thresholdTokens: 272000,
         },
         reasoning: {
             levels: ["none", "low", "medium", "high", "xhigh", "max"],
@@ -891,12 +935,12 @@ export const textModels = [
         inputTokenCost: 0.2,
         cachedInputTokenCost: 0.02,
         outputTokenCost: 1.2,
-        outputTokensPerSecond: 116,
+        outputTokensPerSecond: 165,
         longContext: {
             inputTokenCost: 0.4,
             cachedInputTokenCost: 0.04,
             outputTokenCost: 1.8,
-            thresholdTokens: 200000,
+            thresholdTokens: 272000,
         },
         reasoning: {
             levels: ["none", "low", "medium", "high", "xhigh", "max"],
@@ -927,7 +971,7 @@ export const textModels = [
         inputTokenCost: 2,
         cachedInputTokenCost: 0.2,
         outputTokenCost: 12,
-        outputTokensPerSecond: 133,
+        outputTokensPerSecond: 124,
         longContext: {
             inputTokenCost: 4,
             cachedInputTokenCost: 0.4,
@@ -992,7 +1036,7 @@ export const textModels = [
         inputTokenCost: 0.75,
         cachedInputTokenCost: 0.075,
         outputTokenCost: 3.75,
-        outputTokensPerSecond: 312,
+        outputTokensPerSecond: 329,
         inputAudioTokenCost: 1.5,
         reasoning: {
             levels: ["low", "medium", "high"],
@@ -1424,6 +1468,7 @@ export const textModels = [
         cachedInputTokenCost: 0.25,
         cacheCreationInputTokenCost: 12.5,
         outputTokenCost: 50,
+        outputTokensPerSecond: 69,
         reasoning: {
             thinkingStyle: "adaptive",
             levels: ["low", "medium", "high", "xhigh", "max"],
@@ -1454,6 +1499,7 @@ export const textModels = [
         cachedInputTokenCost: 0.5,
         cacheCreationInputTokenCost: 6.25,
         outputTokenCost: 25,
+        outputTokensPerSecond: 59,
         reasoning: {
             thinkingStyle: "adaptive",
             levels: ["low", "medium", "high", "xhigh", "max"],
@@ -1600,7 +1646,7 @@ export const textModels = [
         cachedInputTokenCost: 0.2,
         cacheCreationInputTokenCost: 2.5,
         outputTokenCost: 10,
-        outputTokensPerSecond: 71,
+        outputTokensPerSecond: 81,
         reasoning: {
             thinkingStyle: "adaptive",
             levels: ["low", "medium", "high", "xhigh", "max"],
@@ -1939,7 +1985,7 @@ export const hostedTools = [
         category: "code_execution",
         description: "Run code in a sandboxed container.",
         providerToolId: "code_execution",
-        pricing: { unit: "per_hour", amount: 0.05, freeAllowance: "50 container-hours/day", note: "Free when used with web_search or web_fetch." },
+        pricing: { unit: "per_hour", amount: 0.05, freeAllowance: "1,550 container-hours/month", note: "Free when used with web_search or web_fetch. 5-minute minimum execution time per container." },
     },
     {
         name: "web_search",
@@ -2014,7 +2060,12 @@ export const hostedTools = [
         description: "Grounding with Google Maps (Gemini 3 only).",
         providerToolId: "google_maps",
         models: ["gemini-3-pro-preview", "gemini-3.1-pro-preview", "gemini-3-flash-preview", "gemini-3.5-flash", "gemini-3.6-flash", "gemini-3.7-flash", "gemini-3.8-flash", "gemini-3.1-flash-lite", "gemini-3.5-flash-lite"],
-        pricing: { unit: "per_call", note: "Gemini 3 family only; see Google pricing." },
+        pricing: {
+            unit: "per_call",
+            amount: 0.014,
+            freeAllowance: "5,000 grounded prompts/month (Gemini 3)",
+            note: "$14 per 1,000 search queries on the Gemini 3 family.",
+        },
     },
     {
         name: "web_search",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "smoltalk",
-  "version": "0.14.1",
+  "version": "0.14.2",
   "description": "A common interface for LLM APIs",
   "homepage": "https://github.com/egonSchiele/smoltalk",
   "files": [