modelfusion 0.98.0 → 0.100.0

package/README.md

@@ -85,7 +85,7 @@ Providers: [OpenAI](https://modelfusion.dev/integration/model-provider/openai),
 Multi-modal vision models such as GPT 4 Vision can process images as part of the prompt.
 
 ```ts
-import { streamText, openai, OpenAIChatMessage } from "modelfusion";
+import { streamText, openai } from "modelfusion";
 import { readFileSync } from "fs";
 
 const image = readFileSync("./image.png").toString("base64");
@@ -93,7 +93,7 @@ const image = readFileSync("./image.png").toString("base64");
 const textStream = await streamText(
   openai.ChatTextGenerator({ model: "gpt-4-vision-preview" }),
   [
-    OpenAIChatMessage.user([
+    openai.ChatMessage.user([
       { type: "text", text: "Describe the image in detail:" },
       { type: "image", base64Image: image, mimeType: "image/png" },
     ]),
@@ -211,7 +211,7 @@ const sentiment = await generateStructure(
     .ChatTextGenerator({
       model: "gpt-3.5-turbo",
       temperature: 0,
-      maxCompletionTokens: 50,
+      maxGenerationTokens: 50,
     })
     .asFunctionCallStructureGenerationModel({ fnName: "sentiment" })
     .withInstructionPrompt(),
@@ -355,14 +355,14 @@ const result = await guard(
   fixStructure({
     modifyInputForRetry: async ({ input, error }) => [
       ...input,
-      OpenAIChatMessage.assistant(null, {
+      openai.ChatMessage.assistant(null, {
         functionCall: {
           name: "sentiment",
           arguments: JSON.stringify(error.valueText),
         },
       }),
-      OpenAIChatMessage.user(error.message),
-      OpenAIChatMessage.user("Please fix the error and try again."),
+      openai.ChatMessage.user(error.message),
+      openai.ChatMessage.user("Please fix the error and try again."),
     ],
   })
 );
@@ -418,7 +418,7 @@ With `generateToolCall`, you can generate a tool call for a specific tool with a
 const { id, name, args } = await generateToolCall(
   openai.ChatTextGenerator({ model: "gpt-3.5-turbo" }),
   calculator,
-  [OpenAIChatMessage.user("What's fourteen times twelve?")]
+  [openai.ChatMessage.user("What's fourteen times twelve?")]
 );
 ```
 
@@ -430,7 +430,7 @@ With `generateToolCallsOrText`, you can ask a language model to generate several
 const { text, toolCalls } = await generateToolCallsOrText(
   openai.ChatTextGenerator({ model: "gpt-3.5-turbo" }),
   [toolA, toolB, toolC],
-  [OpenAIChatMessage.user(query)]
+  [openai.ChatMessage.user(query)]
 );
 ```
 
@@ -454,7 +454,7 @@ With `useTool`, you can use a tool with a language model that supports tools cal
 const { tool, toolCall, args, ok, result } = await useTool(
   openai.ChatTextGenerator({ model: "gpt-3.5-turbo" }),
   calculator,
-  [OpenAIChatMessage.user("What's fourteen times twelve?")]
+  [openai.ChatMessage.user("What's fourteen times twelve?")]
 );
 
 console.log(`Tool call:`, toolCall);
@@ -472,7 +472,7 @@ With `useToolsOrGenerateText`, you can ask a language model to generate several
 const { text, toolResults } = await useToolsOrGenerateText(
   openai.ChatTextGenerator({ model: "gpt-3.5-turbo" }),
   [calculator /* ... */],
-  [OpenAIChatMessage.user("What's fourteen times twelve?")]
+  [openai.ChatMessage.user("What's fourteen times twelve?")]
 );
 ```
 
@@ -541,7 +541,7 @@ const text = await generateText(
   llamacpp
     .TextGenerator({
       contextWindowSize: 4096, // Llama 2 context window size
-      maxCompletionTokens: 1000,
+      maxGenerationTokens: 1000,
     })
     .withTextPromptTemplate(Llama2Prompt.instruction()),
   {
@@ -619,10 +619,10 @@ ModelFusion model functions return rich responses that include the original resp
 
 ```ts
 // access the full response (needs to be typed) and the metadata:
-const { value, response, metadata } = await generateText(
+const { text, response, metadata } = await generateText(
   openai.CompletionTextGenerator({
     model: "gpt-3.5-turbo-instruct",
-    maxCompletionTokens: 1000,
+    maxGenerationTokens: 1000,
     n: 2, // generate 2 completions
   }),
   "Write a short story about a robot learning to love:\n\n",
@@ -767,12 +767,6 @@ Examples for almost all of the individual functions and objects. Highly recommen
 
 StoryTeller is an exploratory web application that creates short audio stories for pre-school kids.
 
-### [Chatbot (Terminal)](https://github.com/lgrammel/modelfusion/tree/main/examples/chatbot-terminal)
-
-> _Terminal app_, _chat_, _llama.cpp_
-
-A chat with an AI assistant, implemented as a terminal app.
-
 ### [Chatbot (Next.JS)](https://github.com/lgrammel/modelfusion/tree/main/examples/chatbot-next-js)
 
 > _Next.js app_, _OpenAI GPT-3.5-turbo_, _streaming_, _abort handling_

package/composed-function/summarize/summarizeRecursivelyWithTextGenerationAndTokenSplitting.cjs

@@ -10,7 +10,7 @@ const summarizeRecursively_js_1 = require("./summarizeRecursively.cjs");
  * while leaving enough space for the model to generate text.
  */
 async function summarizeRecursivelyWithTextGenerationAndTokenSplitting({ text, model, prompt, tokenLimit = model.contextWindowSize -
-    (model.settings.maxCompletionTokens ?? model.contextWindowSize / 4), join, }, options) {
+    (model.settings.maxGenerationTokens ?? model.contextWindowSize / 4), join, }, options) {
     const emptyPromptTokens = await model.countPromptTokens(await prompt({ text: "" }));
     return (0, summarizeRecursively_js_1.summarizeRecursively)({
         split: (0, splitRecursively_js_1.splitAtToken)({

package/composed-function/summarize/summarizeRecursivelyWithTextGenerationAndTokenSplitting.js

@@ -7,7 +7,7 @@ import { summarizeRecursively } from "./summarizeRecursively.js";
  * while leaving enough space for the model to generate text.
  */
 export async function summarizeRecursivelyWithTextGenerationAndTokenSplitting({ text, model, prompt, tokenLimit = model.contextWindowSize -
-    (model.settings.maxCompletionTokens ?? model.contextWindowSize / 4), join, }, options) {
+    (model.settings.maxGenerationTokens ?? model.contextWindowSize / 4), join, }, options) {
     const emptyPromptTokens = await model.countPromptTokens(await prompt({ text: "" }));
     return summarizeRecursively({
         split: splitAtToken({

package/guard/fixStructure.cjs

@@ -45,14 +45,14 @@ const StructureValidationError_js_1 = require("../model-function/generate-struct
  *  fixStructure({
  *    modifyInputForRetry: async ({ input, error }) => [
  *      ...input,
- *      OpenAIChatMessage.assistant(null, {
+ *      openai.ChatMessage.assistant(null, {
  *        functionCall: {
  *          name: "sentiment",
  *          arguments: JSON.stringify(error.valueText),
  *        },
  *      }),
- *      OpenAIChatMessage.user(error.message),
- *      OpenAIChatMessage.user("Please fix the error and try again."),
+ *      openai.ChatMessage.user(error.message),
+ *      openai.ChatMessage.user("Please fix the error and try again."),
  *    ],
  *  })
  * );

package/guard/fixStructure.d.ts

@@ -43,14 +43,14 @@ import { Guard } from "./guard.js";
  *  fixStructure({
  *    modifyInputForRetry: async ({ input, error }) => [
  *      ...input,
- *      OpenAIChatMessage.assistant(null, {
+ *      openai.ChatMessage.assistant(null, {
  *        functionCall: {
  *          name: "sentiment",
  *          arguments: JSON.stringify(error.valueText),
  *        },
  *      }),
- *      OpenAIChatMessage.user(error.message),
- *      OpenAIChatMessage.user("Please fix the error and try again."),
+ *      openai.ChatMessage.user(error.message),
+ *      openai.ChatMessage.user("Please fix the error and try again."),
  *    ],
  *  })
  * );

package/guard/fixStructure.js

@@ -42,14 +42,14 @@ import { StructureValidationError } from "../model-function/generate-structure/S
  *  fixStructure({
  *    modifyInputForRetry: async ({ input, error }) => [
  *      ...input,
- *      OpenAIChatMessage.assistant(null, {
+ *      openai.ChatMessage.assistant(null, {
  *        functionCall: {
  *          name: "sentiment",
  *          arguments: JSON.stringify(error.valueText),
  *        },
  *      }),
- *      OpenAIChatMessage.user(error.message),
- *      OpenAIChatMessage.user("Please fix the error and try again."),
+ *      openai.ChatMessage.user(error.message),
+ *      openai.ChatMessage.user("Please fix the error and try again."),
  *    ],
  *  })
  * );

package/model-function/Model.d.ts

@@ -20,11 +20,11 @@ export interface Model<SETTINGS extends ModelSettings> {
      * @example
      * const model = new OpenAICompletionModel({
      *   model: "gpt-3.5-turbo-instruct",
-     *   maxCompletionTokens: 500,
+     *   maxGenerationTokens: 500,
      * });
      *
      * const modelWithMoreTokens = model.withSettings({
-     *   maxCompletionTokens: 1000,
+     *   maxGenerationTokens: 1000,
      * });
      */
     withSettings(additionalSettings: Partial<SETTINGS>): this;

package/model-function/generate-structure/generateStructure.d.ts

@@ -17,11 +17,11 @@ import { StructureGenerationModel, StructureGenerationModelSettings } from "./St
  *       .describe("Sentiment."),
  *   })),
  *   [
- *     OpenAIChatMessage.system(
+ *     openai.ChatMessage.system(
  *       "You are a sentiment evaluator. " +
  *         "Analyze the sentiment of the following product review:"
  *     ),
- *     OpenAIChatMessage.user(
+ *     openai.ChatMessage.user(
  *       "After I opened the package, I was met by a very unpleasant smell " +
  *         "that did not disappear even after washing. Never again!"
  *     ),

package/model-function/generate-structure/streamStructure.d.ts

@@ -38,7 +38,7 @@ export type StructureStreamPart<STRUCTURE> = {
  *       })
  *     ),
  *   [
- *     OpenAIChatMessage.user(
+ *     openai.ChatMessage.user(
  *       "Generate 3 character descriptions for a fantasy role playing game."
  *     ),
  *   ]

package/model-function/generate-text/PromptTemplateTextGenerationModel.cjs

@@ -40,9 +40,9 @@ class PromptTemplateTextGenerationModel {
         }
         return ((prompt) => originalCountPromptTokens(this.promptTemplate.format(prompt)));
     }
-    doGenerateText(prompt, options) {
+    doGenerateTexts(prompt, options) {
         const mappedPrompt = this.promptTemplate.format(prompt);
-        return this.model.doGenerateText(mappedPrompt, options);
+        return this.model.doGenerateTexts(mappedPrompt, options);
     }
     get settingsForEvent() {
         return this.model.settingsForEvent;

package/model-function/generate-text/PromptTemplateTextGenerationModel.d.ts

@@ -17,9 +17,9 @@ export declare class PromptTemplateTextGenerationModel<PROMPT, MODEL_PROMPT, SET
     get tokenizer(): MODEL["tokenizer"];
     get contextWindowSize(): MODEL["contextWindowSize"];
     get countPromptTokens(): MODEL["countPromptTokens"] extends undefined ? undefined : (prompt: PROMPT) => PromiseLike<number>;
-    doGenerateText(prompt: PROMPT, options?: FunctionOptions): PromiseLike<{
+    doGenerateTexts(prompt: PROMPT, options?: FunctionOptions): PromiseLike<{
         response: unknown;
-        text: string;
+        texts: string[];
         usage?: {
             promptTokens: number;
             completionTokens: number;

package/model-function/generate-text/PromptTemplateTextGenerationModel.js

@@ -37,9 +37,9 @@ export class PromptTemplateTextGenerationModel {
         }
         return ((prompt) => originalCountPromptTokens(this.promptTemplate.format(prompt)));
     }
-    doGenerateText(prompt, options) {
+    doGenerateTexts(prompt, options) {
         const mappedPrompt = this.promptTemplate.format(prompt);
-        return this.model.doGenerateText(mappedPrompt, options);
+        return this.model.doGenerateTexts(mappedPrompt, options);
     }
     get settingsForEvent() {
         return this.model.settingsForEvent;

package/model-function/generate-text/TextGenerationModel.d.ts

@@ -5,18 +5,44 @@ import { BasicTokenizer, FullTokenizer } from "../tokenize-text/Tokenizer.js";
 import { TextGenerationPromptTemplate } from "./TextGenerationPromptTemplate.js";
 export interface TextGenerationModelSettings extends ModelSettings {
     /**
-     * Maximum number of tokens to generate.
+     * Specifies the maximum number of tokens (words, punctuation, parts of words) that the model can generate in a single response.
+     * It helps to control the length of the output.
+     *
      * Does nothing if the model does not support this setting.
+     *
+     * Example: `maxGenerationTokens: 1000`
      */
-    maxCompletionTokens?: number | undefined;
+    maxGenerationTokens?: number | undefined;
     /**
-     * Stop sequences to use. Stop sequences are not included in the generated text.
+     * Stop sequences to use.
+     * Stop sequences are an array of strings or a single string that the model will recognize as end-of-text indicators.
+     * The model stops generating more content when it encounters any of these strings.
+     * This is particularly useful in scripted or formatted text generation, where a specific end point is required.
+     * Stop sequences not included in the generated text.
+     *
      * Does nothing if the model does not support this setting.
+     *
+     * Example: `stopSequences: ['\n', 'END']`
      */
     stopSequences?: string[] | undefined;
+    /**
+     * Number of texts to generate.
+     *
+     * Specifies the number of responses or completions the model should generate for a given prompt.
+     * This is useful when you need multiple different outputs or ideas for a single prompt.
+     * The model will generate 'n' distinct responses, each based on the same initial prompt.
+     * In a streaming model this will result in both responses streamed back in real time.
+     *
+     * Does nothing if the model does not support this setting.
+     *
+     * Example: `numberOfGenerations: 3` // The model will produce 3 different responses.
+     */
+    numberOfGenerations?: number;
     /**
      * When true, the leading and trailing white space and line terminator characters
      * are removed from the generated text.
+     *
+     * Default: true.
      */
     trimWhitespace?: boolean;
 }
@@ -34,9 +60,9 @@ export interface TextGenerationModel<PROMPT, SETTINGS extends TextGenerationMode
      * Optional. Implement if you have a tokenizer and want to count the number of tokens in a prompt.
      */
     readonly countPromptTokens: ((prompt: PROMPT) => PromiseLike<number>) | undefined;
-    doGenerateText(prompt: PROMPT, options?: FunctionOptions): PromiseLike<{
+    doGenerateTexts(prompt: PROMPT, options?: FunctionOptions): PromiseLike<{
         response: unknown;
-        text: string;
+        texts: string[];
         usage?: {
             promptTokens: number;
             completionTokens: number;

package/model-function/generate-text/generateText.cjs

@@ -9,21 +9,27 @@ async function generateText(model, prompt, options) {
         model,
         options,
         generateResponse: async (options) => {
-            const result = await model.doGenerateText(prompt, options);
+            const result = await model.doGenerateTexts(prompt, options);
             const shouldTrimWhitespace = model.settings.trimWhitespace ?? true;
+            const texts = shouldTrimWhitespace
+                ? result.texts.map((text) => text.trim())
+                : result.texts;
             return {
                 response: result.response,
-                extractedValue: shouldTrimWhitespace ? result.text.trim() : result.text,
+                extractedValue: texts,
                 usage: result.usage,
             };
         },
     });
+    const texts = fullResponse.value;
+    const text = texts[0];
     return options?.fullResponse
         ? {
-            text: fullResponse.value,
+            text,
+            texts,
             response: fullResponse.response,
             metadata: fullResponse.metadata,
         }
-        : fullResponse.value;
+        : text;
 }
 exports.generateText = generateText;

package/model-function/generate-text/generateText.d.ts

@@ -29,6 +29,7 @@ export declare function generateText<PROMPT>(model: TextGenerationModel<PROMPT,
     fullResponse: true;
 }): Promise<{
     text: string;
+    texts: string[];
     response: unknown;
     metadata: ModelCallMetadata;
 }>;

package/model-function/generate-text/generateText.js

@@ -6,20 +6,26 @@ export async function generateText(model, prompt, options) {
         model,
         options,
         generateResponse: async (options) => {
-            const result = await model.doGenerateText(prompt, options);
+            const result = await model.doGenerateTexts(prompt, options);
             const shouldTrimWhitespace = model.settings.trimWhitespace ?? true;
+            const texts = shouldTrimWhitespace
+                ? result.texts.map((text) => text.trim())
+                : result.texts;
             return {
                 response: result.response,
-                extractedValue: shouldTrimWhitespace ? result.text.trim() : result.text,
+                extractedValue: texts,
                 usage: result.usage,
             };
         },
     });
+    const texts = fullResponse.value;
+    const text = texts[0];
     return options?.fullResponse
         ? {
-            text: fullResponse.value,
+            text,
+            texts,
             response: fullResponse.response,
             metadata: fullResponse.metadata,
         }
-        : fullResponse.value;
+        : text;
 }

package/model-function/generate-text/prompt-template/trimChatPrompt.cjs

@@ -13,7 +13,7 @@ const ChatPrompt_js_1 = require("./ChatPrompt.cjs");
  * @see https://modelfusion.dev/guide/function/generate-text#limiting-the-chat-length
  */
 async function trimChatPrompt({ prompt, model, tokenLimit = model.contextWindowSize -
-    (model.settings.maxCompletionTokens ?? model.contextWindowSize / 4), }) {
+    (model.settings.maxGenerationTokens ?? model.contextWindowSize / 4), }) {
     (0, ChatPrompt_js_1.validateChatPrompt)(prompt);
     let minimalPrompt = {
         system: prompt.system,

package/model-function/generate-text/prompt-template/trimChatPrompt.js

@@ -10,7 +10,7 @@ import { validateChatPrompt } from "./ChatPrompt.js";
  * @see https://modelfusion.dev/guide/function/generate-text#limiting-the-chat-length
  */
 export async function trimChatPrompt({ prompt, model, tokenLimit = model.contextWindowSize -
-    (model.settings.maxCompletionTokens ?? model.contextWindowSize / 4), }) {
+    (model.settings.maxGenerationTokens ?? model.contextWindowSize / 4), }) {
     validateChatPrompt(prompt);
     let minimalPrompt = {
         system: prompt.system,

package/model-provider/anthropic/AnthropicTextGenerationModel.cjs

@@ -69,22 +69,38 @@ class AnthropicTextGenerationModel extends AbstractModel_js_1.AbstractModel {
         return this.settings.model;
     }
     async callAPI(prompt, options) {
+        const api = this.settings.api ?? new AnthropicApiConfiguration_js_1.AnthropicApiConfiguration();
+        const responseFormat = options.responseFormat;
+        const abortSignal = options.run?.abortSignal;
+        const userId = this.settings.userId;
         return (0, callWithRetryAndThrottle_js_1.callWithRetryAndThrottle)({
             retry: this.settings.api?.retry,
             throttle: this.settings.api?.throttle,
-            call: async () => callAnthropicTextGenerationAPI({
-                ...this.settings,
-                stopSequences: this.settings.stopSequences,
-                maxTokens: this.settings.maxCompletionTokens,
-                abortSignal: options.run?.abortSignal,
-                responseFormat: options.responseFormat,
-                prompt,
-            }),
+            call: async () => {
+                return (0, postToApi_js_1.postJsonToApi)({
+                    url: api.assembleUrl(`/complete`),
+                    headers: api.headers,
+                    body: {
+                        model: this.settings.model,
+                        prompt,
+                        stream: responseFormat.stream,
+                        max_tokens_to_sample: this.settings.maxGenerationTokens,
+                        temperature: this.settings.temperature,
+                        top_k: this.settings.topK,
+                        top_p: this.settings.topP,
+                        stop_sequences: this.settings.stopSequences,
+                        metadata: userId != null ? { user_id: userId } : undefined,
+                    },
+                    failedResponseHandler: AnthropicError_js_1.failedAnthropicCallResponseHandler,
+                    successfulResponseHandler: responseFormat.handler,
+                    abortSignal,
+                });
+            },
         });
     }
     get settingsForEvent() {
         const eventSettingProperties = [
-            "maxCompletionTokens",
+            "maxGenerationTokens",
             "stopSequences",
             "temperature",
             "topK",
@@ -93,14 +109,14 @@ class AnthropicTextGenerationModel extends AbstractModel_js_1.AbstractModel {
         ];
         return Object.fromEntries(Object.entries(this.settings).filter(([key]) => eventSettingProperties.includes(key)));
     }
-    async doGenerateText(prompt, options) {
+    async doGenerateTexts(prompt, options) {
         const response = await this.callAPI(prompt, {
             ...options,
             responseFormat: exports.AnthropicTextGenerationResponseFormat.json,
         });
         return {
             response,
-            text: response.completion,
+            texts: [response.completion],
         };
     }
     doStreamText(prompt, options) {
@@ -148,26 +164,6 @@ const anthropicTextGenerationResponseSchema = zod_1.z.object({
     stop_reason: zod_1.z.string(),
     model: zod_1.z.string(),
 });
-async function callAnthropicTextGenerationAPI({ api = new AnthropicApiConfiguration_js_1.AnthropicApiConfiguration(), abortSignal, responseFormat, model, prompt, maxTokens, stopSequences, temperature, topK, topP, userId, }) {
-    return (0, postToApi_js_1.postJsonToApi)({
-        url: api.assembleUrl(`/complete`),
-        headers: api.headers,
-        body: {
-            model,
-            prompt,
-            stream: responseFormat.stream,
-            max_tokens_to_sample: maxTokens,
-            temperature,
-            top_k: topK,
-            top_p: topP,
-            stop_sequences: stopSequences,
-            metadata: userId != null ? { user_id: userId } : undefined,
-        },
-        failedResponseHandler: AnthropicError_js_1.failedAnthropicCallResponseHandler,
-        successfulResponseHandler: responseFormat.handler,
-        abortSignal,
-    });
-}
 const anthropicTextStreamingResponseSchema = new ZodSchema_js_1.ZodSchema(zod_1.z.object({
     completion: zod_1.z.string(),
     stop_reason: zod_1.z.string().nullable(),

package/model-provider/anthropic/AnthropicTextGenerationModel.d.ts

@@ -49,13 +49,13 @@ export declare class AnthropicTextGenerationModel extends AbstractModel<Anthropi
         responseFormat: AnthropicTextGenerationResponseFormatType<RESPONSE>;
     } & FunctionOptions): Promise<RESPONSE>;
     get settingsForEvent(): Partial<AnthropicTextGenerationModelSettings>;
-    doGenerateText(prompt: string, options?: FunctionOptions): Promise<{
+    doGenerateTexts(prompt: string, options?: FunctionOptions): Promise<{
         response: {
             model: string;
             completion: string;
             stop_reason: string;
         };
-        text: string;
+        texts: string[];
     }>;
     doStreamText(prompt: string, options?: FunctionOptions): Promise<AsyncIterable<Delta<string>>>;
     /**

package/model-provider/anthropic/AnthropicTextGenerationModel.js

@@ -66,22 +66,38 @@ export class AnthropicTextGenerationModel extends AbstractModel {
         return this.settings.model;
     }
     async callAPI(prompt, options) {
+        const api = this.settings.api ?? new AnthropicApiConfiguration();
+        const responseFormat = options.responseFormat;
+        const abortSignal = options.run?.abortSignal;
+        const userId = this.settings.userId;
         return callWithRetryAndThrottle({
             retry: this.settings.api?.retry,
             throttle: this.settings.api?.throttle,
-            call: async () => callAnthropicTextGenerationAPI({
-                ...this.settings,
-                stopSequences: this.settings.stopSequences,
-                maxTokens: this.settings.maxCompletionTokens,
-                abortSignal: options.run?.abortSignal,
-                responseFormat: options.responseFormat,
-                prompt,
-            }),
+            call: async () => {
+                return postJsonToApi({
+                    url: api.assembleUrl(`/complete`),
+                    headers: api.headers,
+                    body: {
+                        model: this.settings.model,
+                        prompt,
+                        stream: responseFormat.stream,
+                        max_tokens_to_sample: this.settings.maxGenerationTokens,
+                        temperature: this.settings.temperature,
+                        top_k: this.settings.topK,
+                        top_p: this.settings.topP,
+                        stop_sequences: this.settings.stopSequences,
+                        metadata: userId != null ? { user_id: userId } : undefined,
+                    },
+                    failedResponseHandler: failedAnthropicCallResponseHandler,
+                    successfulResponseHandler: responseFormat.handler,
+                    abortSignal,
+                });
+            },
         });
     }
     get settingsForEvent() {
         const eventSettingProperties = [
-            "maxCompletionTokens",
+            "maxGenerationTokens",
             "stopSequences",
             "temperature",
             "topK",
@@ -90,14 +106,14 @@ export class AnthropicTextGenerationModel extends AbstractModel {
         ];
         return Object.fromEntries(Object.entries(this.settings).filter(([key]) => eventSettingProperties.includes(key)));
     }
-    async doGenerateText(prompt, options) {
+    async doGenerateTexts(prompt, options) {
         const response = await this.callAPI(prompt, {
             ...options,
             responseFormat: AnthropicTextGenerationResponseFormat.json,
         });
         return {
             response,
-            text: response.completion,
+            texts: [response.completion],
         };
     }
     doStreamText(prompt, options) {
@@ -144,26 +160,6 @@ const anthropicTextGenerationResponseSchema = z.object({
     stop_reason: z.string(),
     model: z.string(),
 });
-async function callAnthropicTextGenerationAPI({ api = new AnthropicApiConfiguration(), abortSignal, responseFormat, model, prompt, maxTokens, stopSequences, temperature, topK, topP, userId, }) {
-    return postJsonToApi({
-        url: api.assembleUrl(`/complete`),
-        headers: api.headers,
-        body: {
-            model,
-            prompt,
-            stream: responseFormat.stream,
-            max_tokens_to_sample: maxTokens,
-            temperature,
-            top_k: topK,
-            top_p: topP,
-            stop_sequences: stopSequences,
-            metadata: userId != null ? { user_id: userId } : undefined,
-        },
-        failedResponseHandler: failedAnthropicCallResponseHandler,
-        successfulResponseHandler: responseFormat.handler,
-        abortSignal,
-    });
-}
 const anthropicTextStreamingResponseSchema = new ZodSchema(z.object({
     completion: z.string(),
     stop_reason: z.string().nullable(),

package/model-provider/cohere/CohereFacade.cjs

@@ -13,7 +13,7 @@ const CohereTokenizer_js_1 = require("./CohereTokenizer.cjs");
  * const model = cohere.TextGenerator({
  *   model: "command-nightly",
  *   temperature: 0.7,
- *   maxCompletionTokens: 500,
+ *   maxGenerationTokens: 500,
  * });
  *
  * const text = await generateText(

package/model-provider/cohere/CohereFacade.d.ts

@@ -10,7 +10,7 @@ import { CohereTokenizer, CohereTokenizerSettings } from "./CohereTokenizer.js";
  * const model = cohere.TextGenerator({
  *   model: "command-nightly",
  *   temperature: 0.7,
- *   maxCompletionTokens: 500,
+ *   maxGenerationTokens: 500,
  * });
  *
  * const text = await generateText(

package/model-provider/cohere/CohereFacade.js

@@ -10,7 +10,7 @@ import { CohereTokenizer } from "./CohereTokenizer.js";
  * const model = cohere.TextGenerator({
  *   model: "command-nightly",
  *   temperature: 0.7,
- *   maxCompletionTokens: 500,
+ *   maxGenerationTokens: 500,
  * });
  *
  * const text = await generateText(