smoltalk 0.0.21 → 0.0.23

package/dist/clients/baseClient.d.ts

@@ -10,6 +10,7 @@ export declare class BaseClient implements SmolClient {
         stream: true;
     }): AsyncGenerator<StreamChunk>;
     text(promptConfig: PromptConfig): Promise<Result<PromptResult>> | AsyncGenerator<StreamChunk>;
+    checkMessageLimit(promptConfig: PromptConfig): Result<PromptResult> | null;
     textSync(promptConfig: PromptConfig): Promise<Result<PromptResult>>;
     checkForToolLoops(promptConfig: PromptConfig): {
         continue: boolean;

package/dist/clients/baseClient.js

@@ -14,7 +14,22 @@ export class BaseClient {
             return this.textSync(promptConfig);
         }
     }
+    checkMessageLimit(promptConfig) {
+        if (promptConfig.maxMessages !== undefined &&
+            promptConfig.messages.length > promptConfig.maxMessages) {
+            const logger = getLogger();
+            logger.warn(`Message limit exceeded: ${promptConfig.messages.length} messages sent, but maxMessages is set to ${promptConfig.maxMessages}. Aborting request.`);
+            return {
+                success: false,
+                error: `Message limit exceeded: ${promptConfig.messages.length} messages exceeds the maxMessages limit of ${promptConfig.maxMessages}`,
+            };
+        }
+        return null;
+    }
     async textSync(promptConfig) {
+        const messageLimitResult = this.checkMessageLimit(promptConfig);
+        if (messageLimitResult)
+            return messageLimitResult;
         const { continue: shouldContinue, newPromptConfig } = this.checkForToolLoops(promptConfig);
         if (!shouldContinue) {
             return {
@@ -97,6 +112,14 @@ export class BaseClient {
         return this.text(newPromptConfig);
     }
     async *textStream(config) {
+        const messageLimitResult = this.checkMessageLimit(config);
+        if (messageLimitResult) {
+            yield {
+                type: "error",
+                error: messageLimitResult.success === false ? messageLimitResult.error : "Message limit exceeded",
+            };
+            return;
+        }
         const { continue: shouldContinue, newPromptConfig } = this.checkForToolLoops(config);
         if (!shouldContinue) {
             yield {

package/dist/models.d.ts

@@ -92,10 +92,19 @@ export declare const textModels: readonly [{
     readonly maxInputTokens: 200000;
     readonly maxOutputTokens: 100000;
     readonly inputTokenCost: 1.1;
-    readonly cachedInputTokenCost: 0.55;
+    readonly cachedInputTokenCost: 0.275;
     readonly outputTokenCost: 4.4;
     readonly outputTokensPerSecond: 135;
     readonly provider: "openai";
+}, {
+    readonly type: "text";
+    readonly modelName: "o3-pro";
+    readonly description: "o3-pro uses more compute for complex reasoning tasks. Available via Responses API only. Requests may take several minutes. Knowledge cutoff: June 2024.";
+    readonly maxInputTokens: 200000;
+    readonly maxOutputTokens: 100000;
+    readonly inputTokenCost: 20;
+    readonly outputTokenCost: 80;
+    readonly provider: "openai";
 }, {
     readonly type: "text";
     readonly modelName: "o1";
@@ -140,17 +149,50 @@ export declare const textModels: readonly [{
 }, {
     readonly type: "text";
     readonly modelName: "gpt-4.1";
-    readonly description: "GPT-4.1 supports up to 1 million tokens of context, representing a significant increase in context window capacity. Ideal for processing large documents and extended conversations.";
+    readonly description: "GPT-4.1 excels at instruction following and tool calling with 1M token context window. Knowledge cutoff: June 2024.";
     readonly maxInputTokens: 1047576;
     readonly maxOutputTokens: 32768;
-    readonly inputTokenCost: 2.5;
-    readonly cachedInputTokenCost: 1.25;
-    readonly outputTokenCost: 10;
+    readonly inputTokenCost: 2;
+    readonly cachedInputTokenCost: 0.5;
+    readonly outputTokenCost: 8;
+    readonly outputTokensPerSecond: 105;
+    readonly provider: "openai";
+}, {
+    readonly type: "text";
+    readonly modelName: "gpt-4.1-mini";
+    readonly description: "GPT-4.1 mini excels at instruction following and tool calling with 1M token context window and low latency. Knowledge cutoff: June 2024.";
+    readonly maxInputTokens: 1047576;
+    readonly maxOutputTokens: 32768;
+    readonly inputTokenCost: 0.4;
+    readonly cachedInputTokenCost: 0.1;
+    readonly outputTokenCost: 1.6;
+    readonly outputTokensPerSecond: 78;
+    readonly provider: "openai";
+}, {
+    readonly type: "text";
+    readonly modelName: "gpt-4.1-nano";
+    readonly description: "GPT-4.1 nano is the fastest and most affordable GPT-4.1 variant with 1M token context window. Knowledge cutoff: June 2024.";
+    readonly maxInputTokens: 1047576;
+    readonly maxOutputTokens: 32768;
+    readonly inputTokenCost: 0.1;
+    readonly cachedInputTokenCost: 0.025;
+    readonly outputTokenCost: 0.4;
+    readonly outputTokensPerSecond: 142;
     readonly provider: "openai";
+}, {
+    readonly type: "text";
+    readonly modelName: "gemini-3.1-pro-preview";
+    readonly description: "Latest Gemini 3.1 Pro with 1M context window and 64K output. Standard pricing for ≤200k tokens ($2.00 input/$12.00 output), higher rates for >200k tokens ($4.00 input/$18.00 output). Released Feb 2026.";
+    readonly maxInputTokens: 1048576;
+    readonly maxOutputTokens: 65536;
+    readonly inputTokenCost: 2;
+    readonly outputTokenCost: 12;
+    readonly outputTokensPerSecond: 112;
+    readonly provider: "google";
 }, {
     readonly type: "text";
     readonly modelName: "gemini-3-pro-preview";
-    readonly description: "Strongest Gemini 3 model quality with 1M context window and 64K output. Standard pricing for ≤200k tokens ($2.00 input/$12.00 output), higher rates for >200k tokens ($4.00 input/$18.00 output). Released Nov 2025, currently in preview.";
+    readonly description: "Gemini 3 Pro with 1M context window and 64K output. Standard pricing for ≤200k tokens ($2.00 input/$12.00 output), higher rates for >200k tokens ($4.00 input/$18.00 output). Superseded by 3.1 Pro.";
     readonly maxInputTokens: 1048576;
     readonly maxOutputTokens: 65536;
     readonly inputTokenCost: 2;
@@ -168,9 +210,9 @@ export declare const textModels: readonly [{
 }, {
     readonly type: "text";
     readonly modelName: "gemini-2.5-pro";
-    readonly description: "High-performance Gemini 2.5 model with 2M context window. Adaptive thinking for complex reasoning and coding. Standard pricing for ≤200k tokens ($1.25 input/$10.00 output), higher rates for >200k tokens ($2.50 input/higher output). Batch API: 50% discount.";
+    readonly description: "High-performance Gemini 2.5 model with 2M context window. Adaptive thinking for complex reasoning and coding. Standard pricing for ≤200k tokens ($1.25 input/$10.00 output), higher rates for >200k tokens ($2.50 input/$15.00 output). Batch API: 50% discount.";
     readonly maxInputTokens: 2097152;
-    readonly maxOutputTokens: 8192;
+    readonly maxOutputTokens: 65536;
     readonly inputTokenCost: 1.25;
     readonly outputTokenCost: 10;
     readonly outputTokensPerSecond: 175;
@@ -180,7 +222,7 @@ export declare const textModels: readonly [{
     readonly modelName: "gemini-2.5-flash";
     readonly description: "Balanced Gemini 2.5 model with excellent performance-to-cost ratio. Lightning-fast with controllable thinking budgets. 1M context window. Context caching available for up to 75% cost reduction.";
     readonly maxInputTokens: 1048576;
-    readonly maxOutputTokens: 8192;
+    readonly maxOutputTokens: 65536;
     readonly inputTokenCost: 0.3;
     readonly outputTokenCost: 2.5;
     readonly outputTokensPerSecond: 225;
@@ -190,7 +232,7 @@ export declare const textModels: readonly [{
     readonly modelName: "gemini-2.5-flash-lite";
     readonly description: "Most cost-effective Gemini 2.5 option for high-throughput applications. 1M context window.";
     readonly maxInputTokens: 1048576;
-    readonly maxOutputTokens: 8192;
+    readonly maxOutputTokens: 65536;
     readonly inputTokenCost: 0.1;
     readonly outputTokenCost: 0.4;
     readonly outputTokensPerSecond: 400;
@@ -219,43 +261,47 @@ export declare const textModels: readonly [{
 }, {
     readonly type: "text";
     readonly modelName: "gemini-2.0-flash-lite";
-    readonly description: "Cost effective offering to support high throughput. Note: May be deprecated in favor of 2.5-flash-lite.";
+    readonly description: "Cost effective offering to support high throughput. DEPRECATED: Will be shut down on March 31, 2026. Use gemini-2.5-flash-lite instead.";
     readonly maxInputTokens: 1048576;
     readonly maxOutputTokens: 8192;
     readonly inputTokenCost: 0.075;
     readonly outputTokenCost: 0.3;
+    readonly disabled: true;
     readonly provider: "google";
 }, {
     readonly type: "text";
     readonly modelName: "gemini-1.5-flash";
-    readonly description: "Provides speed and efficiency for high-volume, quality, cost-effective apps. Note: prices ~double after the first 128k tokens.";
+    readonly description: "RETIRED: No longer available. Use gemini-2.5-flash instead.";
     readonly maxInputTokens: 1048576;
     readonly maxOutputTokens: 8192;
     readonly inputTokenCost: 0.01875;
     readonly outputTokenCost: 0.075;
     readonly outputTokensPerSecond: 178;
     readonly costUnit: "characters";
+    readonly disabled: true;
     readonly provider: "google";
 }, {
     readonly type: "text";
     readonly modelName: "gemini-1.5-pro";
-    readonly description: "Supports text or chat prompts for a text or code response. Supports long-context understanding up to the maximum input token limit. Also does video?";
+    readonly description: "RETIRED: No longer available. Use gemini-2.5-pro instead.";
     readonly maxInputTokens: 2097152;
     readonly maxOutputTokens: 8192;
     readonly inputTokenCost: 0.3125;
     readonly outputTokenCost: 1.25;
     readonly outputTokensPerSecond: 59;
     readonly costUnit: "characters";
+    readonly disabled: true;
     readonly provider: "google";
 }, {
     readonly type: "text";
     readonly modelName: "gemini-1.0-pro";
-    readonly description: "The best performing model for a wide range of text-only tasks.";
+    readonly description: "RETIRED: No longer available. Use gemini-2.5-flash instead.";
     readonly maxInputTokens: 32760;
     readonly maxOutputTokens: 8192;
     readonly inputTokenCost: 0.125;
     readonly outputTokenCost: 0.375;
     readonly costUnit: "characters";
+    readonly disabled: true;
     readonly provider: "google";
 }, {
     readonly type: "text";
@@ -409,10 +455,19 @@ export declare function getModel(modelName: ModelName): {
     readonly maxInputTokens: 200000;
     readonly maxOutputTokens: 100000;
     readonly inputTokenCost: 1.1;
-    readonly cachedInputTokenCost: 0.55;
+    readonly cachedInputTokenCost: 0.275;
     readonly outputTokenCost: 4.4;
     readonly outputTokensPerSecond: 135;
     readonly provider: "openai";
+} | {
+    readonly type: "text";
+    readonly modelName: "o3-pro";
+    readonly description: "o3-pro uses more compute for complex reasoning tasks. Available via Responses API only. Requests may take several minutes. Knowledge cutoff: June 2024.";
+    readonly maxInputTokens: 200000;
+    readonly maxOutputTokens: 100000;
+    readonly inputTokenCost: 20;
+    readonly outputTokenCost: 80;
+    readonly provider: "openai";
 } | {
     readonly type: "text";
     readonly modelName: "o1";
@@ -457,17 +512,50 @@ export declare function getModel(modelName: ModelName): {
 } | {
     readonly type: "text";
     readonly modelName: "gpt-4.1";
-    readonly description: "GPT-4.1 supports up to 1 million tokens of context, representing a significant increase in context window capacity. Ideal for processing large documents and extended conversations.";
+    readonly description: "GPT-4.1 excels at instruction following and tool calling with 1M token context window. Knowledge cutoff: June 2024.";
     readonly maxInputTokens: 1047576;
     readonly maxOutputTokens: 32768;
-    readonly inputTokenCost: 2.5;
-    readonly cachedInputTokenCost: 1.25;
-    readonly outputTokenCost: 10;
+    readonly inputTokenCost: 2;
+    readonly cachedInputTokenCost: 0.5;
+    readonly outputTokenCost: 8;
+    readonly outputTokensPerSecond: 105;
+    readonly provider: "openai";
+} | {
+    readonly type: "text";
+    readonly modelName: "gpt-4.1-mini";
+    readonly description: "GPT-4.1 mini excels at instruction following and tool calling with 1M token context window and low latency. Knowledge cutoff: June 2024.";
+    readonly maxInputTokens: 1047576;
+    readonly maxOutputTokens: 32768;
+    readonly inputTokenCost: 0.4;
+    readonly cachedInputTokenCost: 0.1;
+    readonly outputTokenCost: 1.6;
+    readonly outputTokensPerSecond: 78;
+    readonly provider: "openai";
+} | {
+    readonly type: "text";
+    readonly modelName: "gpt-4.1-nano";
+    readonly description: "GPT-4.1 nano is the fastest and most affordable GPT-4.1 variant with 1M token context window. Knowledge cutoff: June 2024.";
+    readonly maxInputTokens: 1047576;
+    readonly maxOutputTokens: 32768;
+    readonly inputTokenCost: 0.1;
+    readonly cachedInputTokenCost: 0.025;
+    readonly outputTokenCost: 0.4;
+    readonly outputTokensPerSecond: 142;
     readonly provider: "openai";
+} | {
+    readonly type: "text";
+    readonly modelName: "gemini-3.1-pro-preview";
+    readonly description: "Latest Gemini 3.1 Pro with 1M context window and 64K output. Standard pricing for ≤200k tokens ($2.00 input/$12.00 output), higher rates for >200k tokens ($4.00 input/$18.00 output). Released Feb 2026.";
+    readonly maxInputTokens: 1048576;
+    readonly maxOutputTokens: 65536;
+    readonly inputTokenCost: 2;
+    readonly outputTokenCost: 12;
+    readonly outputTokensPerSecond: 112;
+    readonly provider: "google";
 } | {
     readonly type: "text";
     readonly modelName: "gemini-3-pro-preview";
-    readonly description: "Strongest Gemini 3 model quality with 1M context window and 64K output. Standard pricing for ≤200k tokens ($2.00 input/$12.00 output), higher rates for >200k tokens ($4.00 input/$18.00 output). Released Nov 2025, currently in preview.";
+    readonly description: "Gemini 3 Pro with 1M context window and 64K output. Standard pricing for ≤200k tokens ($2.00 input/$12.00 output), higher rates for >200k tokens ($4.00 input/$18.00 output). Superseded by 3.1 Pro.";
     readonly maxInputTokens: 1048576;
     readonly maxOutputTokens: 65536;
     readonly inputTokenCost: 2;
@@ -485,9 +573,9 @@ export declare function getModel(modelName: ModelName): {
 } | {
     readonly type: "text";
     readonly modelName: "gemini-2.5-pro";
-    readonly description: "High-performance Gemini 2.5 model with 2M context window. Adaptive thinking for complex reasoning and coding. Standard pricing for ≤200k tokens ($1.25 input/$10.00 output), higher rates for >200k tokens ($2.50 input/higher output). Batch API: 50% discount.";
+    readonly description: "High-performance Gemini 2.5 model with 2M context window. Adaptive thinking for complex reasoning and coding. Standard pricing for ≤200k tokens ($1.25 input/$10.00 output), higher rates for >200k tokens ($2.50 input/$15.00 output). Batch API: 50% discount.";
     readonly maxInputTokens: 2097152;
-    readonly maxOutputTokens: 8192;
+    readonly maxOutputTokens: 65536;
     readonly inputTokenCost: 1.25;
     readonly outputTokenCost: 10;
     readonly outputTokensPerSecond: 175;
@@ -497,7 +585,7 @@ export declare function getModel(modelName: ModelName): {
     readonly modelName: "gemini-2.5-flash";
     readonly description: "Balanced Gemini 2.5 model with excellent performance-to-cost ratio. Lightning-fast with controllable thinking budgets. 1M context window. Context caching available for up to 75% cost reduction.";
     readonly maxInputTokens: 1048576;
-    readonly maxOutputTokens: 8192;
+    readonly maxOutputTokens: 65536;
     readonly inputTokenCost: 0.3;
     readonly outputTokenCost: 2.5;
     readonly outputTokensPerSecond: 225;
@@ -507,7 +595,7 @@ export declare function getModel(modelName: ModelName): {
     readonly modelName: "gemini-2.5-flash-lite";
     readonly description: "Most cost-effective Gemini 2.5 option for high-throughput applications. 1M context window.";
     readonly maxInputTokens: 1048576;
-    readonly maxOutputTokens: 8192;
+    readonly maxOutputTokens: 65536;
     readonly inputTokenCost: 0.1;
     readonly outputTokenCost: 0.4;
     readonly outputTokensPerSecond: 400;
@@ -536,43 +624,47 @@ export declare function getModel(modelName: ModelName): {
 } | {
     readonly type: "text";
     readonly modelName: "gemini-2.0-flash-lite";
-    readonly description: "Cost effective offering to support high throughput. Note: May be deprecated in favor of 2.5-flash-lite.";
+    readonly description: "Cost effective offering to support high throughput. DEPRECATED: Will be shut down on March 31, 2026. Use gemini-2.5-flash-lite instead.";
     readonly maxInputTokens: 1048576;
     readonly maxOutputTokens: 8192;
     readonly inputTokenCost: 0.075;
     readonly outputTokenCost: 0.3;
+    readonly disabled: true;
     readonly provider: "google";
 } | {
     readonly type: "text";
     readonly modelName: "gemini-1.5-flash";
-    readonly description: "Provides speed and efficiency for high-volume, quality, cost-effective apps. Note: prices ~double after the first 128k tokens.";
+    readonly description: "RETIRED: No longer available. Use gemini-2.5-flash instead.";
     readonly maxInputTokens: 1048576;
     readonly maxOutputTokens: 8192;
     readonly inputTokenCost: 0.01875;
     readonly outputTokenCost: 0.075;
     readonly outputTokensPerSecond: 178;
     readonly costUnit: "characters";
+    readonly disabled: true;
     readonly provider: "google";
 } | {
     readonly type: "text";
     readonly modelName: "gemini-1.5-pro";
-    readonly description: "Supports text or chat prompts for a text or code response. Supports long-context understanding up to the maximum input token limit. Also does video?";
+    readonly description: "RETIRED: No longer available. Use gemini-2.5-pro instead.";
     readonly maxInputTokens: 2097152;
     readonly maxOutputTokens: 8192;
     readonly inputTokenCost: 0.3125;
     readonly outputTokenCost: 1.25;
     readonly outputTokensPerSecond: 59;
     readonly costUnit: "characters";
+    readonly disabled: true;
     readonly provider: "google";
 } | {
     readonly type: "text";
     readonly modelName: "gemini-1.0-pro";
-    readonly description: "The best performing model for a wide range of text-only tasks.";
+    readonly description: "RETIRED: No longer available. Use gemini-2.5-flash instead.";
     readonly maxInputTokens: 32760;
     readonly maxOutputTokens: 8192;
     readonly inputTokenCost: 0.125;
     readonly outputTokenCost: 0.375;
     readonly costUnit: "characters";
+    readonly disabled: true;
     readonly provider: "google";
 } | {
     readonly type: "text";
@@ -663,9 +755,12 @@ export type Optimization = "speed" | "accuracy" | "cost" | "large-context";
 export type ModelConfig = {
     optimizeFor: Optimization[];
     providers: Provider[];
+    limit?: {
+        cost?: number;
+    };
 };
 export declare function isModelConfig(model: ModelName | ModelConfig): model is ModelConfig;
-export declare function pickModel(config: ModelConfig, models?: readonly (typeof textModels)[number][]): TextModelName;
+export declare function pickModel(config: ModelConfig, models?: readonly TextModel[]): TextModelName;
 export declare function calculateCost(modelName: ModelName, usage: {
     inputTokens: number;
     outputTokens: number;

package/dist/models.js

@@ -75,11 +75,21 @@ export const textModels = [
         maxInputTokens: 200000,
         maxOutputTokens: 100000,
         inputTokenCost: 1.1,
-        cachedInputTokenCost: 0.55,
+        cachedInputTokenCost: 0.275,
         outputTokenCost: 4.4,
         outputTokensPerSecond: 135,
         provider: "openai",
     },
+    {
+        type: "text",
+        modelName: "o3-pro",
+        description: "o3-pro uses more compute for complex reasoning tasks. Available via Responses API only. Requests may take several minutes. Knowledge cutoff: June 2024.",
+        maxInputTokens: 200000,
+        maxOutputTokens: 100000,
+        inputTokenCost: 20,
+        outputTokenCost: 80,
+        provider: "openai",
+    },
     {
         type: "text",
         modelName: "o1",
@@ -128,18 +138,54 @@ export const textModels = [
     {
         type: "text",
         modelName: "gpt-4.1",
-        description: "GPT-4.1 supports up to 1 million tokens of context, representing a significant increase in context window capacity. Ideal for processing large documents and extended conversations.",
+        description: "GPT-4.1 excels at instruction following and tool calling with 1M token context window. Knowledge cutoff: June 2024.",
         maxInputTokens: 1047576,
         maxOutputTokens: 32768,
-        inputTokenCost: 2.5,
-        cachedInputTokenCost: 1.25,
-        outputTokenCost: 10,
+        inputTokenCost: 2.0,
+        cachedInputTokenCost: 0.5,
+        outputTokenCost: 8,
+        outputTokensPerSecond: 105,
+        provider: "openai",
+    },
+    {
+        type: "text",
+        modelName: "gpt-4.1-mini",
+        description: "GPT-4.1 mini excels at instruction following and tool calling with 1M token context window and low latency. Knowledge cutoff: June 2024.",
+        maxInputTokens: 1047576,
+        maxOutputTokens: 32768,
+        inputTokenCost: 0.4,
+        cachedInputTokenCost: 0.1,
+        outputTokenCost: 1.6,
+        outputTokensPerSecond: 78,
         provider: "openai",
     },
+    {
+        type: "text",
+        modelName: "gpt-4.1-nano",
+        description: "GPT-4.1 nano is the fastest and most affordable GPT-4.1 variant with 1M token context window. Knowledge cutoff: June 2024.",
+        maxInputTokens: 1047576,
+        maxOutputTokens: 32768,
+        inputTokenCost: 0.1,
+        cachedInputTokenCost: 0.025,
+        outputTokenCost: 0.4,
+        outputTokensPerSecond: 142,
+        provider: "openai",
+    },
+    {
+        type: "text",
+        modelName: "gemini-3.1-pro-preview",
+        description: "Latest Gemini 3.1 Pro with 1M context window and 64K output. Standard pricing for ≤200k tokens ($2.00 input/$12.00 output), higher rates for >200k tokens ($4.00 input/$18.00 output). Released Feb 2026.",
+        maxInputTokens: 1_048_576,
+        maxOutputTokens: 65536,
+        inputTokenCost: 2.0,
+        outputTokenCost: 12.0,
+        outputTokensPerSecond: 112,
+        provider: "google",
+    },
     {
         type: "text",
         modelName: "gemini-3-pro-preview",
-        description: "Strongest Gemini 3 model quality with 1M context window and 64K output. Standard pricing for ≤200k tokens ($2.00 input/$12.00 output), higher rates for >200k tokens ($4.00 input/$18.00 output). Released Nov 2025, currently in preview.",
+        description: "Gemini 3 Pro with 1M context window and 64K output. Standard pricing for ≤200k tokens ($2.00 input/$12.00 output), higher rates for >200k tokens ($4.00 input/$18.00 output). Superseded by 3.1 Pro.",
         maxInputTokens: 1_048_576,
         maxOutputTokens: 65536,
         inputTokenCost: 2.0,
@@ -159,9 +205,9 @@ export const textModels = [
     {
         type: "text",
         modelName: "gemini-2.5-pro",
-        description: "High-performance Gemini 2.5 model with 2M context window. Adaptive thinking for complex reasoning and coding. Standard pricing for ≤200k tokens ($1.25 input/$10.00 output), higher rates for >200k tokens ($2.50 input/higher output). Batch API: 50% discount.",
+        description: "High-performance Gemini 2.5 model with 2M context window. Adaptive thinking for complex reasoning and coding. Standard pricing for ≤200k tokens ($1.25 input/$10.00 output), higher rates for >200k tokens ($2.50 input/$15.00 output). Batch API: 50% discount.",
         maxInputTokens: 2_097_152,
-        maxOutputTokens: 8192,
+        maxOutputTokens: 65536,
         inputTokenCost: 1.25,
         outputTokenCost: 10.0,
         outputTokensPerSecond: 175,
@@ -172,7 +218,7 @@ export const textModels = [
         modelName: "gemini-2.5-flash",
         description: "Balanced Gemini 2.5 model with excellent performance-to-cost ratio. Lightning-fast with controllable thinking budgets. 1M context window. Context caching available for up to 75% cost reduction.",
         maxInputTokens: 1_048_576,
-        maxOutputTokens: 8192,
+        maxOutputTokens: 65536,
         inputTokenCost: 0.3,
         outputTokenCost: 2.5,
         outputTokensPerSecond: 225,
@@ -183,7 +229,7 @@ export const textModels = [
         modelName: "gemini-2.5-flash-lite",
         description: "Most cost-effective Gemini 2.5 option for high-throughput applications. 1M context window.",
         maxInputTokens: 1_048_576,
-        maxOutputTokens: 8192,
+        maxOutputTokens: 65536,
         inputTokenCost: 0.1,
         outputTokenCost: 0.4,
         outputTokensPerSecond: 400,
@@ -215,46 +261,50 @@ export const textModels = [
     {
         type: "text",
         modelName: "gemini-2.0-flash-lite",
-        description: "Cost effective offering to support high throughput. Note: May be deprecated in favor of 2.5-flash-lite.",
+        description: "Cost effective offering to support high throughput. DEPRECATED: Will be shut down on March 31, 2026. Use gemini-2.5-flash-lite instead.",
         maxInputTokens: 1_048_576,
         maxOutputTokens: 8192,
         inputTokenCost: 0.075,
         outputTokenCost: 0.3,
+        disabled: true,
         provider: "google",
     },
     {
         type: "text",
         modelName: "gemini-1.5-flash",
-        description: "Provides speed and efficiency for high-volume, quality, cost-effective apps. Note: prices ~double after the first 128k tokens.",
+        description: "RETIRED: No longer available. Use gemini-2.5-flash instead.",
         maxInputTokens: 1_048_576,
         maxOutputTokens: 8192,
         inputTokenCost: 0.01875,
         outputTokenCost: 0.075,
         outputTokensPerSecond: 178,
         costUnit: "characters",
+        disabled: true,
         provider: "google",
     },
     {
         type: "text",
         modelName: "gemini-1.5-pro",
-        description: "Supports text or chat prompts for a text or code response. Supports long-context understanding up to the maximum input token limit. Also does video?",
+        description: "RETIRED: No longer available. Use gemini-2.5-pro instead.",
         maxInputTokens: 2_097_152,
         maxOutputTokens: 8192,
         inputTokenCost: 0.3125,
         outputTokenCost: 1.25,
         outputTokensPerSecond: 59,
         costUnit: "characters",
+        disabled: true,
         provider: "google",
     },
     {
         type: "text",
         modelName: "gemini-1.0-pro",
-        description: "The best performing model for a wide range of text-only tasks.",
+        description: "RETIRED: No longer available. Use gemini-2.5-flash instead.",
         maxInputTokens: 32_760,
         maxOutputTokens: 8192,
         inputTokenCost: 0.125,
         outputTokenCost: 0.375,
         costUnit: "characters",
+        disabled: true,
         provider: "google",
     },
     {
@@ -429,8 +479,14 @@ function isLowerBetter(optimization) {
     return optimization === "cost";
 }
 export function pickModel(config, models = textModels) {
-    const candidates = models.filter((m) => config.providers.includes(m.provider) &&
+    let candidates = models.filter((m) => config.providers.includes(m.provider) &&
         !("disabled" in m && m.disabled));
+    if (config.limit?.cost !== undefined) {
+        candidates = candidates.filter((m) => {
+            const cost = (m.inputTokenCost ?? 0) + (m.outputTokenCost ?? 0);
+            return cost <= config.limit.cost;
+        });
+    }
     if (candidates.length === 0) {
         throw new SmolError("No models available for providers: " +
             config.providers.join(", ") +

package/dist/types.d.ts

@@ -25,6 +25,7 @@ export type PromptConfig = {
         numRetries: number;
     }>;
     rawAttributes?: Record<string, any>;
+    maxMessages?: number;
 };
 export type SmolConfig = {
     openAiApiKey?: string;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "smoltalk",
-  "version": "0.0.21",
+  "version": "0.0.23",
   "description": "A common interface for LLM APIs",
   "homepage": "https://github.com/egonSchiele/smoltalk",
   "scripts": {