modelfusion 0.135.1 → 0.137.0

package/index.js

@@ -69,6 +69,11 @@ var DefaultRun = class {
       this.functionEventSource.notify(event);
     }
   };
+  getSuccessfulModelCalls() {
+    return this.events.filter(
+      (event) => "model" in event && "result" in event && "status" in event.result && event.result.status === "success"
+    );
+  }
 };
 
 // src/core/ModelFusionConfiguration.ts
@@ -8185,134 +8190,45 @@ async function countOpenAIChatPromptTokens({
 }
 
 // src/model-provider/openai/OpenAIChatModel.ts
-var OPENAI_CHAT_MODELS = {
-  "gpt-4": {
-    contextWindowSize: 8192,
-    promptTokenCostInMillicents: 3,
-    completionTokenCostInMillicents: 6
-  },
-  "gpt-4-0314": {
-    contextWindowSize: 8192,
-    promptTokenCostInMillicents: 3,
-    completionTokenCostInMillicents: 6
-  },
-  "gpt-4-0613": {
-    contextWindowSize: 8192,
-    promptTokenCostInMillicents: 3,
-    completionTokenCostInMillicents: 6,
-    fineTunedPromptTokenCostInMillicents: null,
-    fineTunedCompletionTokenCostInMillicents: null
-  },
-  "gpt-4-turbo-preview": {
-    contextWindowSize: 128e3,
-    promptTokenCostInMillicents: 1,
-    completionTokenCostInMillicents: 3
-  },
-  "gpt-4-1106-preview": {
-    contextWindowSize: 128e3,
-    promptTokenCostInMillicents: 1,
-    completionTokenCostInMillicents: 3
-  },
-  "gpt-4-0125-preview": {
-    contextWindowSize: 128e3,
-    promptTokenCostInMillicents: 1,
-    completionTokenCostInMillicents: 3
-  },
-  "gpt-4-vision-preview": {
-    contextWindowSize: 128e3,
-    promptTokenCostInMillicents: 1,
-    completionTokenCostInMillicents: 3
-  },
-  "gpt-4-32k": {
-    contextWindowSize: 32768,
-    promptTokenCostInMillicents: 6,
-    completionTokenCostInMillicents: 12
-  },
-  "gpt-4-32k-0314": {
-    contextWindowSize: 32768,
-    promptTokenCostInMillicents: 6,
-    completionTokenCostInMillicents: 12
-  },
-  "gpt-4-32k-0613": {
-    contextWindowSize: 32768,
-    promptTokenCostInMillicents: 6,
-    completionTokenCostInMillicents: 12
-  },
-  "gpt-3.5-turbo": {
-    contextWindowSize: 4096,
-    promptTokenCostInMillicents: 0.15,
-    completionTokenCostInMillicents: 0.2,
-    fineTunedPromptTokenCostInMillicents: 0.3,
-    fineTunedCompletionTokenCostInMillicents: 0.6
-  },
-  "gpt-3.5-turbo-0125": {
-    contextWindowSize: 16385,
-    promptTokenCostInMillicents: 0.05,
-    completionTokenCostInMillicents: 0.15
-  },
-  "gpt-3.5-turbo-1106": {
-    contextWindowSize: 16385,
-    promptTokenCostInMillicents: 0.1,
-    completionTokenCostInMillicents: 0.2
-  },
-  "gpt-3.5-turbo-0301": {
-    contextWindowSize: 4096,
-    promptTokenCostInMillicents: 0.15,
-    completionTokenCostInMillicents: 0.2
-  },
-  "gpt-3.5-turbo-0613": {
-    contextWindowSize: 4096,
-    promptTokenCostInMillicents: 0.15,
-    completionTokenCostInMillicents: 0.2,
-    fineTunedPromptTokenCostInMillicents: 1.2,
-    fineTunedCompletionTokenCostInMillicents: 1.6
-  },
-  "gpt-3.5-turbo-16k": {
-    contextWindowSize: 16384,
-    promptTokenCostInMillicents: 0.3,
-    completionTokenCostInMillicents: 0.4
-  },
-  "gpt-3.5-turbo-16k-0613": {
-    contextWindowSize: 16384,
-    promptTokenCostInMillicents: 0.3,
-    completionTokenCostInMillicents: 0.4
-  }
+var CHAT_MODEL_CONTEXT_WINDOW_SIZES = {
+  "gpt-4": 8192,
+  "gpt-4-0314": 8192,
+  "gpt-4-0613": 8192,
+  "gpt-4-turbo-preview": 128e3,
+  "gpt-4-1106-preview": 128e3,
+  "gpt-4-0125-preview": 128e3,
+  "gpt-4-vision-preview": 128e3,
+  "gpt-4-32k": 32768,
+  "gpt-4-32k-0314": 32768,
+  "gpt-4-32k-0613": 32768,
+  "gpt-3.5-turbo": 4096,
+  "gpt-3.5-turbo-0125": 16385,
+  "gpt-3.5-turbo-1106": 16385,
+  "gpt-3.5-turbo-0301": 4096,
+  "gpt-3.5-turbo-0613": 4096,
+  "gpt-3.5-turbo-16k": 16384,
+  "gpt-3.5-turbo-16k-0613": 16384
 };
 function getOpenAIChatModelInformation(model) {
-  if (model in OPENAI_CHAT_MODELS) {
-    const baseModelInformation = OPENAI_CHAT_MODELS[model];
+  if (model in CHAT_MODEL_CONTEXT_WINDOW_SIZES) {
+    const contextWindowSize = CHAT_MODEL_CONTEXT_WINDOW_SIZES[model];
     return {
       baseModel: model,
       isFineTuned: false,
-      contextWindowSize: baseModelInformation.contextWindowSize,
-      promptTokenCostInMillicents: baseModelInformation.promptTokenCostInMillicents,
-      completionTokenCostInMillicents: baseModelInformation.completionTokenCostInMillicents
+      contextWindowSize
     };
   }
   const [_, baseModel, ___, ____, _____] = model.split(":");
   if (["gpt-3.5-turbo", "gpt-3.5-turbo-0613", "gpt-4-0613"].includes(baseModel)) {
-    const baseModelInformation = OPENAI_CHAT_MODELS[baseModel];
+    const contextWindowSize = CHAT_MODEL_CONTEXT_WINDOW_SIZES[baseModel];
     return {
       baseModel,
       isFineTuned: true,
-      contextWindowSize: baseModelInformation.contextWindowSize,
-      promptTokenCostInMillicents: baseModelInformation.fineTunedPromptTokenCostInMillicents,
-      completionTokenCostInMillicents: baseModelInformation.fineTunedCompletionTokenCostInMillicents
+      contextWindowSize
     };
   }
   throw new Error(`Unknown OpenAI chat base model ${baseModel}.`);
 }
-var isOpenAIChatModel = (model) => model in OPENAI_CHAT_MODELS || model.startsWith("ft:gpt-3.5-turbo-0613:") || model.startsWith("ft:gpt-3.5-turbo:");
-var calculateOpenAIChatCostInMillicents = ({
-  model,
-  response
-}) => {
-  const { promptTokenCostInMillicents, completionTokenCostInMillicents } = getOpenAIChatModelInformation(model);
-  if (promptTokenCostInMillicents == null || completionTokenCostInMillicents == null) {
-    return null;
-  }
-  return response.usage.prompt_tokens * promptTokenCostInMillicents + response.usage.completion_tokens * completionTokenCostInMillicents;
-};
 var OpenAIChatModel = class _OpenAIChatModel extends AbstractOpenAIChatModel {
   constructor(settings) {
     super(settings);
@@ -8410,22 +8326,12 @@ var OpenAIChatModel = class _OpenAIChatModel extends AbstractOpenAIChatModel {
 // src/model-provider/openai/OpenAICompletionModel.ts
 var OPENAI_TEXT_GENERATION_MODELS = {
   "gpt-3.5-turbo-instruct": {
-    contextWindowSize: 4097,
-    promptTokenCostInMillicents: 0.15,
-    completionTokenCostInMillicents: 0.2
+    contextWindowSize: 4097
   }
 };
 function getOpenAICompletionModelInformation(model) {
   return OPENAI_TEXT_GENERATION_MODELS[model];
 }
-var isOpenAICompletionModel = (model) => model in OPENAI_TEXT_GENERATION_MODELS;
-var calculateOpenAICompletionCostInMillicents = ({
-  model,
-  response
-}) => {
-  const modelInformation = getOpenAICompletionModelInformation(model);
-  return response.usage.prompt_tokens * modelInformation.promptTokenCostInMillicents + response.usage.completion_tokens * modelInformation.completionTokenCostInMillicents;
-};
 var OpenAICompletionModel = class _OpenAICompletionModel extends AbstractOpenAICompletionModel {
   constructor(settings) {
     super(settings);
@@ -8510,60 +8416,6 @@ __export(OpenAIFacade_exports, {
 
 // src/model-provider/openai/OpenAIImageGenerationModel.ts
 import { z as z27 } from "zod";
-var OPENAI_IMAGE_MODELS = {
-  "dall-e-2": {
-    getCost(settings) {
-      switch (settings.size ?? "1024x1024") {
-        case "1024x1024":
-          return 2e3;
-        case "512x512":
-          return 1800;
-        case "256x256":
-          return 1600;
-        default:
-          return null;
-      }
-    }
-  },
-  "dall-e-3": {
-    getCost(settings) {
-      switch (settings.quality ?? "standard") {
-        case "standard": {
-          switch (settings.size ?? "1024x1024") {
-            case "1024x1024":
-              return 4e3;
-            case "1024x1792":
-            case "1792x1024":
-              return 8e3;
-            default:
-              return null;
-          }
-        }
-        case "hd": {
-          switch (settings.size ?? "1024x1024") {
-            case "1024x1024":
-              return 8e3;
-            case "1024x1792":
-            case "1792x1024":
-              return 12e3;
-            default:
-              return null;
-          }
-        }
-      }
-    }
-  }
-};
-var calculateOpenAIImageGenerationCostInMillicents = ({
-  model,
-  settings
-}) => {
-  const cost = OPENAI_IMAGE_MODELS[model]?.getCost(settings);
-  if (cost == null) {
-    return null;
-  }
-  return (settings.numberOfGenerations ?? 1) * cost;
-};
 var OpenAIImageGenerationModel = class _OpenAIImageGenerationModel extends AbstractModel {
   constructor(settings) {
     super({ settings });
@@ -8667,25 +8519,6 @@ var OpenAIImageGenerationResponseFormat = {
 };
 
 // src/model-provider/openai/OpenAISpeechModel.ts
-var OPENAI_SPEECH_MODELS = {
-  "tts-1": {
-    costInMillicentsPerCharacter: 1.5
-    // = 1500 / 1000,
-  },
-  "tts-1-hd": {
-    costInMillicentsPerCharacter: 3
-    // = 3000 / 1000
-  }
-};
-var calculateOpenAISpeechCostInMillicents = ({
-  model,
-  input
-}) => {
-  if (!OPENAI_SPEECH_MODELS[model]) {
-    return null;
-  }
-  return input.length * OPENAI_SPEECH_MODELS[model].costInMillicentsPerCharacter;
-};
 var OpenAISpeechModel = class _OpenAISpeechModel extends AbstractModel {
   constructor(settings) {
     super({ settings });
@@ -8744,34 +8577,37 @@ var OpenAISpeechModel = class _OpenAISpeechModel extends AbstractModel {
 };
 
 // src/model-provider/openai/OpenAITextEmbeddingModel.ts
+import z28 from "zod";
 var OPENAI_TEXT_EMBEDDING_MODELS = {
   "text-embedding-3-small": {
     contextWindowSize: 8192,
-    dimensions: 1536,
-    tokenCostInMillicents: 2e-3
+    dimensions: 1536
   },
   "text-embedding-3-large": {
     contextWindowSize: 8192,
-    dimensions: 3072,
-    tokenCostInMillicents: 0.013
+    dimensions: 3072
   },
   "text-embedding-ada-002": {
     contextWindowSize: 8192,
-    dimensions: 1536,
-    tokenCostInMillicents: 0.01
-  }
-};
-var isOpenAIEmbeddingModel = (model) => model in OPENAI_TEXT_EMBEDDING_MODELS;
-var calculateOpenAIEmbeddingCostInMillicents = ({
-  model,
-  responses
-}) => {
-  let amountInMilliseconds = 0;
-  for (const response of responses) {
-    amountInMilliseconds += response.usage.total_tokens * OPENAI_TEXT_EMBEDDING_MODELS[model].tokenCostInMillicents;
+    dimensions: 1536
   }
-  return amountInMilliseconds;
 };
+var openAITextEmbeddingResponseSchema2 = z28.object({
+  object: z28.literal("list"),
+  data: z28.array(
+    z28.object({
+      object: z28.literal("embedding"),
+      embedding: z28.array(z28.number()),
+      index: z28.number()
+    })
+  ),
+  model: z28.string(),
+  usage: z28.object({
+    prompt_tokens: z28.number(),
+    total_tokens: z28.number()
+  }).optional()
+  // for openai-compatible models
+});
 var OpenAITextEmbeddingModel = class _OpenAITextEmbeddingModel extends AbstractOpenAITextEmbeddingModel {
   constructor(settings) {
     super(settings);
@@ -8800,7 +8636,7 @@ var OpenAITextEmbeddingModel = class _OpenAITextEmbeddingModel extends AbstractO
 };
 
 // src/model-provider/openai/OpenAITranscriptionModel.ts
-import { z as z28 } from "zod";
+import { z as z29 } from "zod";
 
 // src/util/audio/getAudioFileExtension.ts
 function getAudioFileExtension(mimeType) {
@@ -8830,22 +8666,6 @@ function getAudioFileExtension(mimeType) {
 }
 
 // src/model-provider/openai/OpenAITranscriptionModel.ts
-var OPENAI_TRANSCRIPTION_MODELS = {
-  "whisper-1": {
-    costInMillicentsPerSecond: 10
-    // = 600 / 60,
-  }
-};
-var calculateOpenAITranscriptionCostInMillicents = ({
-  model,
-  response
-}) => {
-  if (model !== "whisper-1") {
-    return null;
-  }
-  const durationInSeconds = response.duration;
-  return Math.ceil(durationInSeconds) * OPENAI_TRANSCRIPTION_MODELS[model].costInMillicentsPerSecond;
-};
 var OpenAITranscriptionModel = class _OpenAITranscriptionModel extends AbstractModel {
   constructor(settings) {
     super({ settings });
@@ -8931,29 +8751,29 @@ var OpenAITranscriptionModel = class _OpenAITranscriptionModel extends AbstractM
     );
   }
 };
-var openAITranscriptionJsonSchema = z28.object({
-  text: z28.string()
+var openAITranscriptionJsonSchema = z29.object({
+  text: z29.string()
 });
-var openAITranscriptionVerboseJsonSchema = z28.object({
-  task: z28.literal("transcribe"),
-  language: z28.string(),
-  duration: z28.number(),
-  segments: z28.array(
-    z28.object({
-      id: z28.number(),
-      seek: z28.number(),
-      start: z28.number(),
-      end: z28.number(),
-      text: z28.string(),
-      tokens: z28.array(z28.number()),
-      temperature: z28.number(),
-      avg_logprob: z28.number(),
-      compression_ratio: z28.number(),
-      no_speech_prob: z28.number(),
-      transient: z28.boolean().optional()
+var openAITranscriptionVerboseJsonSchema = z29.object({
+  task: z29.literal("transcribe"),
+  language: z29.string(),
+  duration: z29.number(),
+  segments: z29.array(
+    z29.object({
+      id: z29.number(),
+      seek: z29.number(),
+      start: z29.number(),
+      end: z29.number(),
+      text: z29.string(),
+      tokens: z29.array(z29.number()),
+      temperature: z29.number(),
+      avg_logprob: z29.number(),
+      compression_ratio: z29.number(),
+      no_speech_prob: z29.number(),
+      transient: z29.boolean().optional()
     })
   ),
-  text: z28.string()
+  text: z29.string()
 });
 var OpenAITranscriptionResponseFormat = {
   json: {
@@ -9293,9 +9113,9 @@ var StabilityApiConfiguration = class extends BaseUrlApiConfigurationWithDefault
 };
 
 // src/model-provider/stability/StabilityError.ts
-import { z as z29 } from "zod";
-var stabilityErrorDataSchema = z29.object({
-  message: z29.string()
+import { z as z30 } from "zod";
+var stabilityErrorDataSchema = z30.object({
+  message: z30.string()
 });
 var failedStabilityCallResponseHandler = createJsonErrorResponseHandler({
   errorSchema: zodSchema(stabilityErrorDataSchema),
@@ -9310,7 +9130,7 @@ __export(StabilityFacade_exports, {
 });
 
 // src/model-provider/stability/StabilityImageGenerationModel.ts
-import { z as z30 } from "zod";
+import { z as z31 } from "zod";
 
 // src/model-provider/stability/StabilityImageGenerationPrompt.ts
 function mapBasicPromptToStabilityFormat() {
@@ -9399,12 +9219,12 @@ var StabilityImageGenerationModel = class _StabilityImageGenerationModel extends
     );
   }
 };
-var stabilityImageGenerationResponseSchema = z30.object({
-  artifacts: z30.array(
-    z30.object({
-      base64: z30.string(),
-      seed: z30.number(),
-      finishReason: z30.enum(["SUCCESS", "ERROR", "CONTENT_FILTERED"])
+var stabilityImageGenerationResponseSchema = z31.object({
+  artifacts: z31.array(
+    z31.object({
+      base64: z31.string(),
+      seed: z31.number(),
+      finishReason: z31.enum(["SUCCESS", "ERROR", "CONTENT_FILTERED"])
     })
   )
 });
@@ -9440,7 +9260,7 @@ __export(WhisperCppFacade_exports, {
 });
 
 // src/model-provider/whispercpp/WhisperCppTranscriptionModel.ts
-import { z as z31 } from "zod";
+import { z as z32 } from "zod";
 var WhisperCppTranscriptionModel = class _WhisperCppTranscriptionModel extends AbstractModel {
   constructor(settings) {
     super({ settings });
@@ -9511,9 +9331,9 @@ var WhisperCppTranscriptionModel = class _WhisperCppTranscriptionModel extends A
     );
   }
 };
-var whisperCppTranscriptionJsonSchema = z31.union([
-  z31.object({ text: z31.string() }),
-  z31.object({ error: z31.string() })
+var whisperCppTranscriptionJsonSchema = z32.union([
+  z32.object({ text: z32.string() }),
+  z32.object({ error: z32.string() })
 ]);
 var successfulResponseHandler = async ({ response, url, requestBodyValues }) => {
   const responseBody = await response.text();
@@ -9872,14 +9692,14 @@ var ToolExecutionError = class extends Error {
 };
 
 // src/tool/WebSearchTool.ts
-import { z as z32 } from "zod";
+import { z as z33 } from "zod";
 var RETURN_TYPE_SCHEMA = zodSchema(
-  z32.object({
-    results: z32.array(
-      z32.object({
-        title: z32.string(),
-        link: z32.string().url(),
-        snippet: z32.string()
+  z33.object({
+    results: z33.array(
+      z33.object({
+        title: z33.string(),
+        link: z33.string().url(),
+        snippet: z33.string()
       })
     )
   })
@@ -9887,8 +9707,8 @@ var RETURN_TYPE_SCHEMA = zodSchema(
 var createParameters = (description) => (
   // same schema, but with description:
   zodSchema(
-    z32.object({
-      query: z32.string().describe(description)
+    z33.object({
+      query: z33.string().describe(description)
     })
   )
 );
@@ -10362,13 +10182,13 @@ var VectorIndexRetriever = class _VectorIndexRetriever {
 };
 
 // src/vector-index/memory/MemoryVectorIndex.ts
-import { z as z33 } from "zod";
+import { z as z34 } from "zod";
 var jsonDataSchema = zodSchema(
-  z33.array(
-    z33.object({
-      id: z33.string(),
-      vector: z33.array(z33.number()),
-      data: z33.unknown()
+  z34.array(
+    z34.object({
+      id: z34.string(),
+      vector: z34.array(z34.number()),
+      data: z34.unknown()
     })
   )
 );
@@ -10466,6 +10286,7 @@ export {
   AzureOpenAIApiConfiguration,
   BaseUrlApiConfiguration,
   BaseUrlApiConfigurationWithDefaults,
+  CHAT_MODEL_CONTEXT_WINDOW_SIZES,
   COHERE_TEXT_EMBEDDING_MODELS,
   COHERE_TEXT_GENERATION_MODELS,
   ChatMLPromptTemplate_exports as ChatMLPrompt,
@@ -10505,13 +10326,9 @@ export {
   NeuralChatPromptTemplate_exports as NeuralChatPrompt,
   NoSuchToolDefinitionError,
   OPENAI_CHAT_MESSAGE_BASE_TOKEN_COUNT,
-  OPENAI_CHAT_MODELS,
   OPENAI_CHAT_PROMPT_BASE_TOKEN_COUNT,
-  OPENAI_IMAGE_MODELS,
-  OPENAI_SPEECH_MODELS,
   OPENAI_TEXT_EMBEDDING_MODELS,
   OPENAI_TEXT_GENERATION_MODELS,
-  OPENAI_TRANSCRIPTION_MODELS,
   ObjectFromTextGenerationModel,
   ObjectFromTextStreamingModel,
   ObjectGeneratorTool,
@@ -10571,12 +10388,6 @@ export {
   ZodSchema,
   ApiFacade_exports as api,
   Automatic1111Facade_exports as automatic1111,
-  calculateOpenAIChatCostInMillicents,
-  calculateOpenAICompletionCostInMillicents,
-  calculateOpenAIEmbeddingCostInMillicents,
-  calculateOpenAIImageGenerationCostInMillicents,
-  calculateOpenAISpeechCostInMillicents,
-  calculateOpenAITranscriptionCostInMillicents,
   classify,
   CohereFacade_exports as cohere,
   convertDataContentToBase64String,
@@ -10607,9 +10418,6 @@ export {
   getOpenAICompletionModelInformation,
   getRun,
   HuggingFaceFacade_exports as huggingface,
-  isOpenAIChatModel,
-  isOpenAICompletionModel,
-  isOpenAIEmbeddingModel,
   isPromptFunction,
   jsonObjectPrompt,
   jsonToolCallPrompt,
@@ -10621,6 +10429,7 @@ export {
   MistralFacade_exports as mistral,
   ModelFusionConfiguration_exports as modelfusion,
   OllamaFacade_exports as ollama,
+  openAITextEmbeddingResponseSchema2 as openAITextEmbeddingResponseSchema,
   OpenAIFacade_exports as openai,
   OpenAICompatibleFacade_exports as openaicompatible,
   parseJSON,