modelfusion 0.112.0 → 0.113.0

package/CHANGELOG.md

@@ -1,5 +1,51 @@
 # Changelog
 
+## v0.113.0 - 2024-01-03
+
+[Structure generation](https://modelfusion.dev/guide/function/generate-structure) improvements.
+
+### Added
+
+- `.asStructureGenerationModel(...)` function to `OpenAIChatModel` and `OllamaChatModel` to create structure generation models from chat models.
+- `jsonStructurePrompt` helper function to create structure generation models.
+
+### Example
+
+```ts
+import {
+  generateStructure,
+  jsonStructurePrompt,
+  ollama,
+  zodSchema,
+} from "modelfusion";
+
+const structure = await generateStructure(
+  ollama
+    .ChatTextGenerator({
+      model: "openhermes2.5-mistral",
+      maxGenerationTokens: 1024,
+      temperature: 0,
+    })
+    .asStructureGenerationModel(jsonStructurePrompt.text()),
+
+  zodSchema(
+    z.object({
+      characters: z.array(
+        z.object({
+          name: z.string(),
+          class: z
+            .string()
+            .describe("Character class, e.g. warrior, mage, or thief."),
+          description: z.string(),
+        })
+      ),
+    })
+  ),
+
+  "Generate 3 character descriptions for a fantasy role playing game. "
+);
+```
+
 ## v0.112.0 - 2024-01-02
 
 ### Changed

package/README.md

@@ -1,6 +1,6 @@
 # ModelFusion
 
-> ### The TypeScript library for building multi-modal AI applications.
+> ### The TypeScript library for building sAI applications.
 
 [![NPM Version](https://img.shields.io/npm/v/modelfusion?color=33cd56&logo=npm)](https://www.npmjs.com/package/modelfusion)
 [![MIT License](https://img.shields.io/github/license/lgrammel/modelfusion)](https://opensource.org/licenses/MIT)
@@ -12,7 +12,7 @@
 
 ## Introduction
 
-**ModelFusion** is an abstraction layer for integrating AI models into JavaScript and TypeScript applications, unifying the API for common operations such as text streaming, structure generation, and tool usage. It provides features to support production environments, including observability hooks, logging, and automatic retries. You can use ModelFusion to build AI applications, chatbots, and agents.
+**ModelFusion** is an abstraction layer for integrating AI models into JavaScript and TypeScript applications, unifying the API for common operations such as **text streaming**, **structure generation**, and **tool usage**. It provides features to support production environments, including observability hooks, logging, and automatic retries. You can use ModelFusion to build AI applications, chatbots, and agents.
 
 - **Vendor-neutral**: ModelFusion is a non-commercial open source project that is community-driven. You can use it with any supported provider.
 - **Multi-modal**: ModelFusion supports a wide range of models including text generation, image generation, vision, text-to-speech, speech-to-text, and embedding models.
@@ -201,18 +201,22 @@ Generate typed objects using a language model and a schema.
 Generate a structure that matches a schema.
 
 ```ts
-import { zodSchema, generateStructure, openai } from "modelfusion";
+import {
+  ollama,
+  zodSchema,
+  generateStructure,
+  jsonStructurePrompt,
+} from "modelfusion";
 
 const sentiment = await generateStructure(
   // model:
-  openai
+  ollama
     .ChatTextGenerator({
-      model: "gpt-3.5-turbo",
+      model: "openhermes2.5-mistral",
+      maxGenerationTokens: 1024,
       temperature: 0,
-      maxGenerationTokens: 50,
     })
-    .asFunctionCallStructureGenerationModel({ fnName: "sentiment" })
-    .withInstructionPrompt(),
+    .asStructureGenerationModel(jsonStructurePrompt.instruction()),
 
   // schema:
   zodSchema(
@@ -326,25 +330,13 @@ Providers: [OpenAI](https://modelfusion.dev/integration/model-provider/openai),
 
 ### [Tools](https://modelfusion.dev/guide/tools)
 
-Tools are functions that can be executed by an AI model. They are useful for building chatbots and agents.
+Tools are functions (and associated metadata) that can be executed by an AI model. They are useful for building chatbots and agents.
 
-Predefined tools: [Math.js](https://modelfusion.dev/guide/tools/predefined-tools/mathjs), [MediaWiki Search](https://modelfusion.dev/guide/tools/predefined-tools/mediawiki-search), [SerpAPI](https://modelfusion.dev/guide/tools/predefined-tools/serpapi), [Google Custom Search](https://modelfusion.dev/guide/tools/predefined-tools/google-custom-search)
-
-#### [executeTool](https://modelfusion.dev/guide/tools/execute-tool)
-
-You can directly invoke a tool with `executeTool`:
-
-```ts
-const result = await executeTool(calculator, {
-  a: 14,
-  b: 12,
-  operator: "*",
-});
-```
+ModelFusion offers several tools out-of-the-box: [Math.js](https://modelfusion.dev/guide/tools/available-tools/mathjs), [MediaWiki Search](https://modelfusion.dev/guide/tools/available-tools/mediawiki-search), [SerpAPI](https://modelfusion.dev/guide/tools/available-tools/serpapi), [Google Custom Search](https://modelfusion.dev/guide/tools/available-tools/google-custom-search). You can also create [custom tools](https://modelfusion.dev/guide/tools).
 
 #### [useTool](https://modelfusion.dev/guide/tools/use-tool)
 
-With `useTool`, you can use a tool with a language model that supports tools calls (e.g. OpenAI Chat). `useTool` first generates a tool call and then executes the tool with the arguments.
+With `useTool`, you can ask a tool-compatible language model (e.g. OpenAI chat) to invoke a single tool. `useTool` first generates a tool call and then executes the tool with the arguments.
 
 ```ts
 const { tool, toolCall, args, ok, result } = await useTool(
@@ -360,7 +352,7 @@ console.log(`Ok:`, ok);
 console.log(`Result or Error:`, result);
 ```
 
-#### [useTools](https://modelfusion.dev/guide/tools/use-tools-or-generate-text)
+#### [useTools](https://modelfusion.dev/guide/tools/use-tools)
 
 With `useTools`, you can ask a language model to generate several tool calls as well as text. The model will choose which tools (if any) should be called with which arguments. Both the text and the tool calls are optional. This function executes the tools.
 
@@ -372,42 +364,6 @@ const { text, toolResults } = await useTools(
 );
 ```
 
-#### [Creating Tools](https://modelfusion.dev/guide/tools/create-tools)
-
-A tool is comprised of an async execute function, a name, a description, and a schema for the input parameters.
-
-```ts
-const calculator = new Tool({
-  name: "calculator",
-  description: "Execute a calculation",
-
-  parameters: zodSchema(
-    z.object({
-      a: z.number().describe("The first number."),
-      b: z.number().describe("The second number."),
-      operator: z
-        .enum(["+", "-", "*", "/"])
-        .describe("The operator (+, -, *, /)."),
-    })
-  ),
-
-  execute: async ({ a, b, operator }) => {
-    switch (operator) {
-      case "+":
-        return a + b;
-      case "-":
-        return a - b;
-      case "*":
-        return a * b;
-      case "/":
-        return a / b;
-      default:
-        throw new Error(`Unknown operator: ${operator}`);
-    }
-  },
-});
-```
-
 #### [Agent Loop](https://modelfusion.dev/guide/tools/agent-loop)
 
 You can use `useTools` to implement an agent loop that responds to user messages and executes tools. [Learn more](https://modelfusion.dev/guide/tools/agent-loop).
@@ -582,85 +538,25 @@ import { modelfusion } from "modelfusion";
 modelfusion.setLogFormat("detailed-object"); // log full events
 ```
 
-### [Server](https://modelfusion.dev/guide/experimental/server/)
-
-> [!WARNING]
-> ModelFusion Server is in its initial development phase and not feature-complete. The API is experimental and breaking changes are likely. Feedback and suggestions are welcome.
-
-ModelFusion Server is desigend for running multi-modal generative AI flows that take up to several minutes to complete. It provides the following benefits:
-
-- 🔄 Real-time progress updates via custom server-sent events
-- 🔒Type-safety with Zod-schema for inputs/events
-- 📦 Efficient handling of dynamically created binary assets (images, audio)
-- 📜 Auto-logging for AI model interactions within flows
-
-ModelFusion provides a [Fastify](https://fastify.dev/) plugin that allows you to set up a server that exposes your ModelFusion flows as REST endpoints using server-sent events.
-
-```ts
-import {
-  FileSystemAssetStorage,
-  FileSystemLogger,
-  modelFusionFastifyPlugin,
-} from "modelfusion-experimental/fastify-server"; // '/fastify-server' import path
-
-// configurable logging for all runs using ModelFusion observability:
-const logger = new FileSystemLogger({
-  path: (run) => path.join(fsBasePath, run.runId, "logs"),
-});
-
-// configurable storage for large files like images and audio files:
-const assetStorage = new FileSystemAssetStorage({
-  path: (run) => path.join(fsBasePath, run.runId, "assets"),
-  logger,
-});
-
-fastify.register(modelFusionFastifyPlugin, {
-  baseUrl,
-  basePath: "/myFlow",
-  logger,
-  assetStorage,
-  flow: exampleFlow,
-});
-```
-
-Using `invokeFlow`, you can easily connect your client to a ModelFusion flow endpoint:
-
-```ts
-import { invokeFlow } from "modelfusion-experimental/browser"; // '/browser' import path
-
-invokeFlow({
-  url: `${BASE_URL}/myFlow`,
-  schema: myFlowSchema,
-  input: { prompt },
-  onEvent(event) {
-    switch (event.type) {
-      case "my-event": {
-        // do something with the event
-        break;
-      }
-      // more events...
-    }
-  },
-  onStop() {
-    // flow finished
-  },
-});
-```
-
 ## Documentation
 
 ### [Guide](https://modelfusion.dev/guide)
 
 - [Model Functions](https://modelfusion.dev/guide/function/)
   - [Generate text](https://modelfusion.dev/guide/function/generate-text)
+  - [Generate structure](https://modelfusion.dev/guide/function/generate-structure)
   - [Generate image](https://modelfusion.dev/guide/function/generate-image)
   - [Generate speech](https://modelfusion.dev/guide/function/generate-speech)
   - [Generate transcription](https://modelfusion.dev/guide/function/generation-transcription)
-  - [Generate structure](https://modelfusion.dev/guide/function/generate-structure)
-  - [Generate structure or text](https://modelfusion.dev/guide/function/generate-structure-or-text)
   - [Tokenize Text](https://modelfusion.dev/guide/function/tokenize-text)
   - [Embed Value](https://modelfusion.dev/guide/function/embed)
 - [Tools](https://modelfusion.dev/guide/tools)
+  - [Use Tool](https://modelfusion.dev/guide/tools/use-tool)
+  - [Use Tools](https://modelfusion.dev/guide/tools/use-tools)
+  - [Agent Loop](https://modelfusion.dev/guide/tools/agent-loop)
+  - [Available Tools](https://modelfusion.dev/guide/tools/available-tools/)
+  - [Custom Tools](https://modelfusion.dev/guide/tools/custom-tools)
+  - [Advanced](https://modelfusion.dev/guide/tools/advanced)
 - [Vector Indices](https://modelfusion.dev/guide/vector-index)
   - [Upsert](https://modelfusion.dev/guide/vector-index/upsert)
   - [Retrieve](https://modelfusion.dev/guide/vector-index/retrieve)

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

@@ -1,6 +1,19 @@
 import { JsonSchemaProducer } from "../../core/schema/JsonSchemaProducer.js";
 import { Schema } from "../../core/schema/Schema.js";
+import { TextStreamingModel } from "../generate-text/TextGenerationModel.js";
+import { ChatPrompt } from "../generate-text/prompt-template/ChatPrompt.js";
+import { InstructionPrompt } from "../generate-text/prompt-template/InstructionPrompt.js";
 export type StructureFromTextPromptTemplate<SOURCE_PROMPT, TARGET_PROMPT> = {
     createPrompt: (prompt: SOURCE_PROMPT, schema: Schema<unknown> & JsonSchemaProducer) => TARGET_PROMPT;
     extractStructure: (response: string) => unknown;
 };
+export type FlexibleStructureFromTextPromptTemplate<SOURCE_PROMPT, INTERMEDIATE_PROMPT> = {
+    createPrompt: (prompt: SOURCE_PROMPT, schema: Schema<unknown> & JsonSchemaProducer) => INTERMEDIATE_PROMPT;
+    extractStructure: (response: string) => unknown;
+    adaptModel: (model: TextStreamingModel<never> & {
+        withTextPrompt(): TextStreamingModel<string>;
+        withInstructionPrompt(): TextStreamingModel<InstructionPrompt>;
+        withChatPrompt(): TextStreamingModel<ChatPrompt>;
+        withJsonOutput?: () => typeof model;
+    }) => TextStreamingModel<INTERMEDIATE_PROMPT>;
+};

package/model-function/generate-structure/jsonStructurePrompt.cjs

@@ -15,6 +15,12 @@ exports.jsonStructurePrompt = {
                 instruction: prompt,
             }),
             extractStructure,
+            adaptModel: (model) => {
+                if (model.withJsonOutput != null) {
+                    model = model.withJsonOutput();
+                }
+                return model.withInstructionPrompt();
+            },
         };
     },
     instruction({ schemaPrefix, schemaSuffix, } = {}) {
@@ -29,6 +35,12 @@ exports.jsonStructurePrompt = {
                 instruction: prompt.instruction,
             }),
             extractStructure,
+            adaptModel: (model) => {
+                if (model.withJsonOutput != null) {
+                    model = model.withJsonOutput();
+                }
+                return model.withInstructionPrompt();
+            },
         };
     },
 };

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

@@ -1,15 +1,15 @@
 import { JsonSchemaProducer } from "../../core/schema/JsonSchemaProducer.js";
 import { Schema } from "../../core/schema/Schema.js";
 import { InstructionPrompt } from "../../model-function/generate-text/prompt-template/InstructionPrompt.js";
-import { StructureFromTextPromptTemplate } from "./StructureFromTextPromptTemplate.js";
+import { FlexibleStructureFromTextPromptTemplate, StructureFromTextPromptTemplate } from "./StructureFromTextPromptTemplate.js";
 export declare const jsonStructurePrompt: {
     custom<SOURCE_PROMPT, TARGET_PROMPT>(createPrompt: (prompt: SOURCE_PROMPT, schema: Schema<unknown> & JsonSchemaProducer) => TARGET_PROMPT): StructureFromTextPromptTemplate<SOURCE_PROMPT, TARGET_PROMPT>;
     text({ schemaPrefix, schemaSuffix, }?: {
         schemaPrefix?: string | undefined;
         schemaSuffix?: string | undefined;
-    }): StructureFromTextPromptTemplate<string, InstructionPrompt>;
+    }): FlexibleStructureFromTextPromptTemplate<string, InstructionPrompt>;
     instruction({ schemaPrefix, schemaSuffix, }?: {
         schemaPrefix?: string | undefined;
         schemaSuffix?: string | undefined;
-    }): StructureFromTextPromptTemplate<InstructionPrompt, InstructionPrompt>;
+    }): FlexibleStructureFromTextPromptTemplate<InstructionPrompt, InstructionPrompt>;
 };

package/model-function/generate-structure/jsonStructurePrompt.js

@@ -12,6 +12,12 @@ export const jsonStructurePrompt = {
                 instruction: prompt,
             }),
             extractStructure,
+            adaptModel: (model) => {
+                if (model.withJsonOutput != null) {
+                    model = model.withJsonOutput();
+                }
+                return model.withInstructionPrompt();
+            },
         };
     },
     instruction({ schemaPrefix, schemaSuffix, } = {}) {
@@ -26,6 +32,12 @@ export const jsonStructurePrompt = {
                 instruction: prompt.instruction,
             }),
             extractStructure,
+            adaptModel: (model) => {
+                if (model.withJsonOutput != null) {
+                    model = model.withJsonOutput();
+                }
+                return model.withInstructionPrompt();
+            },
         };
     },
 };

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

@@ -72,6 +72,10 @@ export interface TextGenerationModel<PROMPT, SETTINGS extends TextGenerationMode
         };
     }>;
     withPromptTemplate<INPUT_PROMPT>(promptTemplate: TextGenerationPromptTemplate<INPUT_PROMPT, PROMPT>): TextGenerationModel<INPUT_PROMPT, SETTINGS>;
+    /**
+     * Optional. When available, forces the model to return JSON as the text output.
+     */
+    withJsonOutput?(): this;
 }
 export interface TextStreamingModel<PROMPT, SETTINGS extends TextGenerationModelSettings = TextGenerationModelSettings> extends TextGenerationModel<PROMPT, SETTINGS> {
     doStreamText(prompt: PROMPT, options?: FunctionOptions): PromiseLike<AsyncIterable<Delta<unknown>>>;

package/model-provider/ollama/OllamaChatModel.cjs

@@ -8,6 +8,7 @@ const postToApi_js_1 = require("../../core/api/postToApi.cjs");
 const ZodSchema_js_1 = require("../../core/schema/ZodSchema.cjs");
 const parseJSON_js_1 = require("../../core/schema/parseJSON.cjs");
 const AbstractModel_js_1 = require("../../model-function/AbstractModel.cjs");
+const StructureFromTextStreamingModel_js_1 = require("../../model-function/generate-structure/StructureFromTextStreamingModel.cjs");
 const PromptTemplateTextStreamingModel_js_1 = require("../../model-function/generate-text/PromptTemplateTextStreamingModel.cjs");
 const TextGenerationModel_js_1 = require("../../model-function/generate-text/TextGenerationModel.cjs");
 const TextGenerationToolCallModel_js_1 = require("../../tool/generate-tool-call/TextGenerationToolCallModel.cjs");
@@ -148,6 +149,17 @@ class OllamaChatModel extends AbstractModel_js_1.AbstractModel {
             template: promptTemplate,
         });
     }
+    asStructureGenerationModel(promptTemplate) {
+        return "adaptModel" in promptTemplate
+            ? new StructureFromTextStreamingModel_js_1.StructureFromTextStreamingModel({
+                model: promptTemplate.adaptModel(this),
+                template: promptTemplate,
+            })
+            : new StructureFromTextStreamingModel_js_1.StructureFromTextStreamingModel({
+                model: this,
+                template: promptTemplate,
+            });
+    }
     /**
      * Returns this model with a text prompt template.
      */
@@ -177,6 +189,9 @@ class OllamaChatModel extends AbstractModel_js_1.AbstractModel {
             promptTemplate,
         });
     }
+    withJsonOutput() {
+        return this.withSettings({ format: "json" });
+    }
     withSettings(additionalSettings) {
         return new OllamaChatModel(Object.assign({}, this.settings, additionalSettings));
     }

package/model-provider/ollama/OllamaChatModel.d.ts

@@ -3,6 +3,8 @@ import { FunctionOptions } from "../../core/FunctionOptions.js";
 import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
 import { ResponseHandler } from "../../core/api/postToApi.js";
 import { AbstractModel } from "../../model-function/AbstractModel.js";
+import { FlexibleStructureFromTextPromptTemplate, StructureFromTextPromptTemplate } from "../../model-function/generate-structure/StructureFromTextPromptTemplate.js";
+import { StructureFromTextStreamingModel } from "../../model-function/generate-structure/StructureFromTextStreamingModel.js";
 import { PromptTemplateTextStreamingModel } from "../../model-function/generate-text/PromptTemplateTextStreamingModel.js";
 import { TextStreamingModel } from "../../model-function/generate-text/TextGenerationModel.js";
 import { TextGenerationPromptTemplate } from "../../model-function/generate-text/TextGenerationPromptTemplate.js";
@@ -79,6 +81,7 @@ export declare class OllamaChatModel extends AbstractModel<OllamaChatModelSettin
     extractTextDelta(delta: unknown): string | undefined;
     asToolCallGenerationModel<INPUT_PROMPT>(promptTemplate: ToolCallPromptTemplate<INPUT_PROMPT, OllamaChatPrompt>): TextGenerationToolCallModel<INPUT_PROMPT, OllamaChatPrompt, this>;
     asToolCallsOrTextGenerationModel<INPUT_PROMPT>(promptTemplate: ToolCallsPromptTemplate<INPUT_PROMPT, OllamaChatPrompt>): TextGenerationToolCallsModel<INPUT_PROMPT, OllamaChatPrompt, this>;
+    asStructureGenerationModel<INPUT_PROMPT, OllamaChatPrompt>(promptTemplate: StructureFromTextPromptTemplate<INPUT_PROMPT, OllamaChatPrompt> | FlexibleStructureFromTextPromptTemplate<INPUT_PROMPT, unknown>): StructureFromTextStreamingModel<INPUT_PROMPT, unknown, TextStreamingModel<unknown, import("../../model-function/generate-text/TextGenerationModel.js").TextGenerationModelSettings>> | StructureFromTextStreamingModel<INPUT_PROMPT, OllamaChatPrompt, TextStreamingModel<OllamaChatPrompt, import("../../model-function/generate-text/TextGenerationModel.js").TextGenerationModelSettings>>;
     /**
      * Returns this model with a text prompt template.
      */
@@ -92,6 +95,7 @@ export declare class OllamaChatModel extends AbstractModel<OllamaChatModelSettin
      */
     withChatPrompt(): PromptTemplateTextStreamingModel<import("../../index.js").ChatPrompt, OllamaChatPrompt, OllamaChatModelSettings, this>;
     withPromptTemplate<INPUT_PROMPT>(promptTemplate: TextGenerationPromptTemplate<INPUT_PROMPT, OllamaChatPrompt>): PromptTemplateTextStreamingModel<INPUT_PROMPT, OllamaChatPrompt, OllamaChatModelSettings, this>;
+    withJsonOutput(): this;
     withSettings(additionalSettings: Partial<OllamaChatModelSettings>): this;
 }
 declare const ollamaChatResponseSchema: z.ZodObject<{

package/model-provider/ollama/OllamaChatModel.js

@@ -5,6 +5,7 @@ import { postJsonToApi } from "../../core/api/postToApi.js";
 import { zodSchema } from "../../core/schema/ZodSchema.js";
 import { safeParseJSON } from "../../core/schema/parseJSON.js";
 import { AbstractModel } from "../../model-function/AbstractModel.js";
+import { StructureFromTextStreamingModel } from "../../model-function/generate-structure/StructureFromTextStreamingModel.js";
 import { PromptTemplateTextStreamingModel } from "../../model-function/generate-text/PromptTemplateTextStreamingModel.js";
 import { textGenerationModelProperties, } from "../../model-function/generate-text/TextGenerationModel.js";
 import { TextGenerationToolCallModel, } from "../../tool/generate-tool-call/TextGenerationToolCallModel.js";
@@ -145,6 +146,17 @@ export class OllamaChatModel extends AbstractModel {
             template: promptTemplate,
         });
     }
+    asStructureGenerationModel(promptTemplate) {
+        return "adaptModel" in promptTemplate
+            ? new StructureFromTextStreamingModel({
+                model: promptTemplate.adaptModel(this),
+                template: promptTemplate,
+            })
+            : new StructureFromTextStreamingModel({
+                model: this,
+                template: promptTemplate,
+            });
+    }
     /**
      * Returns this model with a text prompt template.
      */
@@ -174,6 +186,9 @@ export class OllamaChatModel extends AbstractModel {
             promptTemplate,
         });
     }
+    withJsonOutput() {
+        return this.withSettings({ format: "json" });
+    }
     withSettings(additionalSettings) {
         return new OllamaChatModel(Object.assign({}, this.settings, additionalSettings));
     }

package/model-provider/openai/OpenAIChatModel.cjs

@@ -219,10 +219,15 @@ class OpenAIChatModel extends AbstractOpenAIChatModel_js_1.AbstractOpenAIChatMod
         });
     }
     asStructureGenerationModel(promptTemplate) {
-        return new StructureFromTextStreamingModel_js_1.StructureFromTextStreamingModel({
-            model: this,
-            template: promptTemplate,
-        });
+        return "adaptModel" in promptTemplate
+            ? new StructureFromTextStreamingModel_js_1.StructureFromTextStreamingModel({
+                model: promptTemplate.adaptModel(this),
+                template: promptTemplate,
+            })
+            : new StructureFromTextStreamingModel_js_1.StructureFromTextStreamingModel({
+                model: this,
+                template: promptTemplate,
+            });
     }
     /**
      * Returns this model with a text prompt template.
@@ -253,6 +258,9 @@ class OpenAIChatModel extends AbstractOpenAIChatModel_js_1.AbstractOpenAIChatMod
             promptTemplate,
         });
     }
+    withJsonOutput() {
+        return this.withSettings({ responseFormat: { type: "json_object" } });
+    }
     withSettings(additionalSettings) {
         return new OpenAIChatModel(Object.assign({}, this.settings, additionalSettings));
     }

package/model-provider/openai/OpenAIChatModel.d.ts

@@ -1,4 +1,4 @@
-import { StructureFromTextPromptTemplate } from "../../model-function/generate-structure/StructureFromTextPromptTemplate.js";
+import { FlexibleStructureFromTextPromptTemplate, StructureFromTextPromptTemplate } from "../../model-function/generate-structure/StructureFromTextPromptTemplate.js";
 import { StructureFromTextStreamingModel } from "../../model-function/generate-structure/StructureFromTextStreamingModel.js";
 import { PromptTemplateFullTextModel } from "../../model-function/generate-text/PromptTemplateFullTextModel.js";
 import { TextStreamingModel } from "../../model-function/generate-text/TextGenerationModel.js";
@@ -140,7 +140,7 @@ export declare class OpenAIChatModel extends AbstractOpenAIChatModel<OpenAIChatS
         fnName: string;
         fnDescription?: string;
     }): OpenAIChatFunctionCallStructureGenerationModel<TextGenerationPromptTemplate<OpenAIChatPrompt, OpenAIChatPrompt>>;
-    asStructureGenerationModel<INPUT_PROMPT>(promptTemplate: StructureFromTextPromptTemplate<INPUT_PROMPT, OpenAIChatPrompt>): StructureFromTextStreamingModel<INPUT_PROMPT, OpenAIChatPrompt, this>;
+    asStructureGenerationModel<INPUT_PROMPT, OpenAIChatPrompt>(promptTemplate: StructureFromTextPromptTemplate<INPUT_PROMPT, OpenAIChatPrompt> | FlexibleStructureFromTextPromptTemplate<INPUT_PROMPT, unknown>): StructureFromTextStreamingModel<INPUT_PROMPT, unknown, TextStreamingModel<unknown, import("../../model-function/generate-text/TextGenerationModel.js").TextGenerationModelSettings>> | StructureFromTextStreamingModel<INPUT_PROMPT, OpenAIChatPrompt, TextStreamingModel<OpenAIChatPrompt, import("../../model-function/generate-text/TextGenerationModel.js").TextGenerationModelSettings>>;
     /**
      * Returns this model with a text prompt template.
      */
@@ -154,6 +154,7 @@ export declare class OpenAIChatModel extends AbstractOpenAIChatModel<OpenAIChatS
      */
     withChatPrompt(): PromptTemplateFullTextModel<import("../../index.js").ChatPrompt, OpenAIChatPrompt, OpenAIChatSettings, this>;
     withPromptTemplate<INPUT_PROMPT>(promptTemplate: TextGenerationPromptTemplate<INPUT_PROMPT, OpenAIChatPrompt>): PromptTemplateFullTextModel<INPUT_PROMPT, OpenAIChatPrompt, OpenAIChatSettings, this>;
+    withJsonOutput(): this;
     withSettings(additionalSettings: Partial<OpenAIChatSettings>): this;
 }
 export {};

package/model-provider/openai/OpenAIChatModel.js

@@ -213,10 +213,15 @@ export class OpenAIChatModel extends AbstractOpenAIChatModel {
         });
     }
     asStructureGenerationModel(promptTemplate) {
-        return new StructureFromTextStreamingModel({
-            model: this,
-            template: promptTemplate,
-        });
+        return "adaptModel" in promptTemplate
+            ? new StructureFromTextStreamingModel({
+                model: promptTemplate.adaptModel(this),
+                template: promptTemplate,
+            })
+            : new StructureFromTextStreamingModel({
+                model: this,
+                template: promptTemplate,
+            });
     }
     /**
      * Returns this model with a text prompt template.
@@ -247,6 +252,9 @@ export class OpenAIChatModel extends AbstractOpenAIChatModel {
             promptTemplate,
         });
     }
+    withJsonOutput() {
+        return this.withSettings({ responseFormat: { type: "json_object" } });
+    }
     withSettings(additionalSettings) {
         return new OpenAIChatModel(Object.assign({}, this.settings, additionalSettings));
     }

package/model-provider/openai-compatible/OpenAICompatibleChatModel.cjs

@@ -58,10 +58,15 @@ class OpenAICompatibleChatModel extends AbstractOpenAIChatModel_js_1.AbstractOpe
         return Object.fromEntries(Object.entries(this.settings).filter(([key]) => eventSettingProperties.includes(key)));
     }
     asStructureGenerationModel(promptTemplate) {
-        return new StructureFromTextStreamingModel_js_1.StructureFromTextStreamingModel({
-            model: this,
-            template: promptTemplate,
-        });
+        return "adaptModel" in promptTemplate
+            ? new StructureFromTextStreamingModel_js_1.StructureFromTextStreamingModel({
+                model: promptTemplate.adaptModel(this),
+                template: promptTemplate,
+            })
+            : new StructureFromTextStreamingModel_js_1.StructureFromTextStreamingModel({
+                model: this,
+                template: promptTemplate,
+            });
     }
     /**
      * Returns this model with a text prompt template.
@@ -92,6 +97,9 @@ class OpenAICompatibleChatModel extends AbstractOpenAIChatModel_js_1.AbstractOpe
             promptTemplate,
         });
     }
+    withJsonOutput() {
+        return this.withSettings({ responseFormat: { type: "json_object" } });
+    }
     withSettings(additionalSettings) {
         return new OpenAICompatibleChatModel(Object.assign({}, this.settings, additionalSettings));
     }

package/model-provider/openai-compatible/OpenAICompatibleChatModel.d.ts

@@ -1,4 +1,4 @@
-import { StructureFromTextPromptTemplate } from "../../model-function/generate-structure/StructureFromTextPromptTemplate.js";
+import { FlexibleStructureFromTextPromptTemplate, StructureFromTextPromptTemplate } from "../../model-function/generate-structure/StructureFromTextPromptTemplate.js";
 import { StructureFromTextStreamingModel } from "../../model-function/generate-structure/StructureFromTextStreamingModel.js";
 import { PromptTemplateFullTextModel } from "../../model-function/generate-text/PromptTemplateFullTextModel.js";
 import { TextStreamingModel } from "../../model-function/generate-text/TextGenerationModel.js";
@@ -26,7 +26,7 @@ export declare class OpenAICompatibleChatModel extends AbstractOpenAIChatModel<O
     readonly tokenizer: undefined;
     readonly countPromptTokens: undefined;
     get settingsForEvent(): Partial<OpenAICompatibleChatSettings>;
-    asStructureGenerationModel<INPUT_PROMPT>(promptTemplate: StructureFromTextPromptTemplate<INPUT_PROMPT, OpenAIChatPrompt>): StructureFromTextStreamingModel<INPUT_PROMPT, OpenAIChatPrompt, this>;
+    asStructureGenerationModel<INPUT_PROMPT, OpenAIChatPrompt>(promptTemplate: StructureFromTextPromptTemplate<INPUT_PROMPT, OpenAIChatPrompt> | FlexibleStructureFromTextPromptTemplate<INPUT_PROMPT, unknown>): StructureFromTextStreamingModel<INPUT_PROMPT, unknown, TextStreamingModel<unknown, import("../../model-function/generate-text/TextGenerationModel.js").TextGenerationModelSettings>> | StructureFromTextStreamingModel<INPUT_PROMPT, OpenAIChatPrompt, TextStreamingModel<OpenAIChatPrompt, import("../../model-function/generate-text/TextGenerationModel.js").TextGenerationModelSettings>>;
     /**
      * Returns this model with a text prompt template.
      */
@@ -40,5 +40,6 @@ export declare class OpenAICompatibleChatModel extends AbstractOpenAIChatModel<O
      */
     withChatPrompt(): PromptTemplateFullTextModel<import("../../index.js").ChatPrompt, OpenAIChatPrompt, OpenAICompatibleChatSettings, this>;
     withPromptTemplate<INPUT_PROMPT>(promptTemplate: TextGenerationPromptTemplate<INPUT_PROMPT, OpenAIChatPrompt>): PromptTemplateFullTextModel<INPUT_PROMPT, OpenAIChatPrompt, OpenAICompatibleChatSettings, this>;
+    withJsonOutput(): this;
     withSettings(additionalSettings: Partial<OpenAICompatibleChatSettings>): this;
 }

package/model-provider/openai-compatible/OpenAICompatibleChatModel.js

@@ -55,10 +55,15 @@ export class OpenAICompatibleChatModel extends AbstractOpenAIChatModel {
         return Object.fromEntries(Object.entries(this.settings).filter(([key]) => eventSettingProperties.includes(key)));
     }
     asStructureGenerationModel(promptTemplate) {
-        return new StructureFromTextStreamingModel({
-            model: this,
-            template: promptTemplate,
-        });
+        return "adaptModel" in promptTemplate
+            ? new StructureFromTextStreamingModel({
+                model: promptTemplate.adaptModel(this),
+                template: promptTemplate,
+            })
+            : new StructureFromTextStreamingModel({
+                model: this,
+                template: promptTemplate,
+            });
     }
     /**
      * Returns this model with a text prompt template.
@@ -89,6 +94,9 @@ export class OpenAICompatibleChatModel extends AbstractOpenAIChatModel {
             promptTemplate,
         });
     }
+    withJsonOutput() {
+        return this.withSettings({ responseFormat: { type: "json_object" } });
+    }
     withSettings(additionalSettings) {
         return new OpenAICompatibleChatModel(Object.assign({}, this.settings, additionalSettings));
     }

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "modelfusion",
-  "description": "The TypeScript library for building multi-modal AI applications.",
-  "version": "0.112.0",
+  "description": "The TypeScript library for building AI applications.",
+  "version": "0.113.0",
   "author": "Lars Grammel",
   "license": "MIT",
   "keywords": [