🚀 feat: gpt-4o-mini (danny-avila#3384)

* feat: `gpt-4o-mini`

* feat: retrival

* fix: Update order of model token values for 'gpt-4o' and 'gpt-4o-mini'

* fix: Update order of model token values for 'gpt-4o' and 'gpt-4o-mini'

* fix: Update order of model token values for 'gpt-4o' and 'gpt-4o-mini'

* fix: add jsdoc

* fix: Update order of model token values for 'gpt-4o' and 'gpt-4o-mini'

---------

Co-authored-by: Danny Avila <[email protected]>

Author: berry-13

.env.example

@@ -144,7 +144,7 @@ GOOGLE_KEY=user_provided
 #============#
 
 OPENAI_API_KEY=user_provided
-# OPENAI_MODELS=gpt-4o,gpt-3.5-turbo-0125,gpt-3.5-turbo-0301,gpt-3.5-turbo,gpt-4,gpt-4-0613,gpt-4-vision-preview,gpt-3.5-turbo-0613,gpt-3.5-turbo-16k-0613,gpt-4-0125-preview,gpt-4-turbo-preview,gpt-4-1106-preview,gpt-3.5-turbo-1106,gpt-3.5-turbo-instruct,gpt-3.5-turbo-instruct-0914,gpt-3.5-turbo-16k
+# OPENAI_MODELS=gpt-4o,gpt-4o-mini,gpt-3.5-turbo-0125,gpt-3.5-turbo-0301,gpt-3.5-turbo,gpt-4,gpt-4-0613,gpt-4-vision-preview,gpt-3.5-turbo-0613,gpt-3.5-turbo-16k-0613,gpt-4-0125-preview,gpt-4-turbo-preview,gpt-4-1106-preview,gpt-3.5-turbo-1106,gpt-3.5-turbo-instruct,gpt-3.5-turbo-instruct-0914,gpt-3.5-turbo-16k
 
 DEBUG_OPENAI=false
 
@@ -166,7 +166,7 @@ DEBUG_OPENAI=false
 
 ASSISTANTS_API_KEY=user_provided
 # ASSISTANTS_BASE_URL=
-# ASSISTANTS_MODELS=gpt-4o,gpt-3.5-turbo-0125,gpt-3.5-turbo-16k-0613,gpt-3.5-turbo-16k,gpt-3.5-turbo,gpt-4,gpt-4-0314,gpt-4-32k-0314,gpt-4-0613,gpt-3.5-turbo-0613,gpt-3.5-turbo-1106,gpt-4-0125-preview,gpt-4-turbo-preview,gpt-4-1106-preview
+# ASSISTANTS_MODELS=gpt-4o,gpt-4o-mini,gpt-3.5-turbo-0125,gpt-3.5-turbo-16k-0613,gpt-3.5-turbo-16k,gpt-3.5-turbo,gpt-4,gpt-4-0314,gpt-4-32k-0314,gpt-4-0613,gpt-3.5-turbo-0613,gpt-3.5-turbo-1106,gpt-4-0125-preview,gpt-4-turbo-preview,gpt-4-1106-preview
 
 #==========================#
 #   Azure Assistants API   #
@@ -188,7 +188,7 @@ ASSISTANTS_API_KEY=user_provided
 # Plugins    #
 #============#
 
-# PLUGIN_MODELS=gpt-4o,gpt-4,gpt-4-turbo-preview,gpt-4-0125-preview,gpt-4-1106-preview,gpt-4-0613,gpt-3.5-turbo,gpt-3.5-turbo-0125,gpt-3.5-turbo-1106,gpt-3.5-turbo-0613
+# PLUGIN_MODELS=gpt-4o,gpt-4o-mini,gpt-4,gpt-4-turbo-preview,gpt-4-0125-preview,gpt-4-1106-preview,gpt-4-0613,gpt-3.5-turbo,gpt-3.5-turbo-0125,gpt-3.5-turbo-1106,gpt-3.5-turbo-0613
 
 DEBUG_PLUGINS=true
 

README.md

@@ -50,7 +50,7 @@
 - 🔄 Edit, Resubmit, and Continue Messages with Conversation branching
 - 🌿 Fork Messages & Conversations for Advanced Context control
 - 💬 Multimodal Chat:
-    - Upload and analyze images with Claude 3, GPT-4 (including `gpt-4o`), and Gemini Vision 📸
+    - Upload and analyze images with Claude 3, GPT-4 (including `gpt-4o` and `gpt-4o-mini`), and Gemini Vision 📸
     - Chat with Files using Custom Endpoints, OpenAI, Azure, Anthropic, & Google. 🗃️
     - Advanced Agents with Files, Code Interpreter, Tools, and API Actions 🔦
       - Available through the [OpenAI Assistants API](https://platform.openai.com/docs/assistants/overview) 🌤️

api/models/tx.js

@@ -12,6 +12,7 @@ const tokenValues = {
   '4k': { prompt: 1.5, completion: 2 },
   '16k': { prompt: 3, completion: 4 },
   'gpt-3.5-turbo-1106': { prompt: 1, completion: 2 },
+  'gpt-4o-mini': { prompt: 0.15, completion: 0.6 },
   'gpt-4o': { prompt: 5, completion: 15 },
   'gpt-4-1106': { prompt: 10, completion: 30 },
   'gpt-3.5-turbo-0125': { prompt: 0.5, completion: 1.5 },
@@ -54,6 +55,8 @@ const getValueKey = (model, endpoint) => {
     return 'gpt-3.5-turbo-1106';
   } else if (modelName.includes('gpt-3.5')) {
     return '4k';
+  } else if (modelName.includes('gpt-4o-mini')) {
+    return 'gpt-4o-mini';
   } else if (modelName.includes('gpt-4o')) {
     return 'gpt-4o';
   } else if (modelName.includes('gpt-4-vision')) {

api/models/tx.spec.js

@@ -49,6 +49,12 @@ describe('getValueKey', () => {
     expect(getValueKey('gpt-4o-0125')).toBe('gpt-4o');
   });
 
+  it('should return "gpt-4o-mini" for model type of "gpt-4o-mini"', () => {
+    expect(getValueKey('gpt-4o-mini-2024-07-18')).toBe('gpt-4o-mini');
+    expect(getValueKey('openai/gpt-4o-mini')).toBe('gpt-4o-mini');
+    expect(getValueKey('gpt-4o-mini-0718')).toBe('gpt-4o-mini');
+  });
+
   it('should return "claude-3-5-sonnet" for model type of "claude-3-5-sonnet-"', () => {
     expect(getValueKey('claude-3-5-sonnet-20240620')).toBe('claude-3-5-sonnet');
     expect(getValueKey('anthropic/claude-3-5-sonnet')).toBe('claude-3-5-sonnet');
@@ -109,6 +115,19 @@ describe('getMultiplier', () => {
     );
   });
 
+  it('should return the correct multiplier for gpt-4o-mini', () => {
+    const valueKey = getValueKey('gpt-4o-mini-2024-07-18');
+    expect(getMultiplier({ valueKey, tokenType: 'prompt' })).toBe(
+      tokenValues['gpt-4o-mini'].prompt,
+    );
+    expect(getMultiplier({ valueKey, tokenType: 'completion' })).toBe(
+      tokenValues['gpt-4o-mini'].completion,
+    );
+    expect(getMultiplier({ valueKey, tokenType: 'completion' })).not.toBe(
+      tokenValues['gpt-4-1106'].completion,
+    );
+  });
+
   it('should derive the valueKey from the model if not provided for new models', () => {
     expect(
       getMultiplier({ tokenType: 'prompt', model: 'gpt-3.5-turbo-1106-some-other-info' }),

api/utils/tokens.js

@@ -1,45 +1,6 @@
 const z = require('zod');
 const { EModelEndpoint } = require('librechat-data-provider');
 
-const models = [
-  'text-davinci-003',
-  'text-davinci-002',
-  'text-davinci-001',
-  'text-curie-001',
-  'text-babbage-001',
-  'text-ada-001',
-  'davinci',
-  'curie',
-  'babbage',
-  'ada',
-  'code-davinci-002',
-  'code-davinci-001',
-  'code-cushman-002',
-  'code-cushman-001',
-  'davinci-codex',
-  'cushman-codex',
-  'text-davinci-edit-001',
-  'code-davinci-edit-001',
-  'text-embedding-ada-002',
-  'text-similarity-davinci-001',
-  'text-similarity-curie-001',
-  'text-similarity-babbage-001',
-  'text-similarity-ada-001',
-  'text-search-davinci-doc-001',
-  'text-search-curie-doc-001',
-  'text-search-babbage-doc-001',
-  'text-search-ada-doc-001',
-  'code-search-babbage-code-001',
-  'code-search-ada-code-001',
-  'gpt2',
-  'gpt-4',
-  'gpt-4-0314',
-  'gpt-4-32k',
-  'gpt-4-32k-0314',
-  'gpt-3.5-turbo',
-  'gpt-3.5-turbo-0301',
-];
-
 const openAIModels = {
   'gpt-4': 8187, // -5 from max
   'gpt-4-0613': 8187, // -5 from max
@@ -49,6 +10,7 @@ const openAIModels = {
   'gpt-4-1106': 127990, // -10 from max
   'gpt-4-0125': 127990, // -10 from max
   'gpt-4o': 127990, // -10 from max
+  'gpt-4o-mini': 127990, // -10 from max
   'gpt-4-turbo': 127990, // -10 from max
   'gpt-4-vision': 127990, // -10 from max
   'gpt-3.5-turbo': 16375, // -10 from max
@@ -101,7 +63,6 @@ const anthropicModels = {
 
 const aggregateModels = { ...openAIModels, ...googleModels, ...anthropicModels, ...cohereModels };
 
-// Order is important here: by model series and context size (gpt-4 then gpt-3, ascending)
 const maxTokensMap = {
   [EModelEndpoint.azureOpenAI]: openAIModels,
   [EModelEndpoint.openAI]: aggregateModels,
@@ -110,6 +71,24 @@ const maxTokensMap = {
   [EModelEndpoint.anthropic]: anthropicModels,
 };
 
+/**
+ * Finds the first matching pattern in the tokens map.
+ * @param {string} modelName
+ * @param {Record<string, number>} tokensMap
+ * @returns {string|null}
+ */
+function findMatchingPattern(modelName, tokensMap) {
+  const keys = Object.keys(tokensMap);
+  for (let i = keys.length - 1; i >= 0; i--) {
+    const modelKey = keys[i];
+    if (modelName.includes(modelKey)) {
+      return modelKey;
+    }
+  }
+
+  return null;
+}
+
 /**
  * Retrieves the maximum tokens for a given model name. If the exact model name isn't found,
  * it searches for partial matches within the model name, checking keys in reverse order.
@@ -143,12 +122,11 @@ function getModelMaxTokens(modelName, endpoint = EModelEndpoint.openAI, endpoint
     return tokensMap[modelName];
   }
 
-  const keys = Object.keys(tokensMap);
-  for (let i = keys.length - 1; i >= 0; i--) {
-    if (modelName.includes(keys[i])) {
-      const result = tokensMap[keys[i]];
-      return result?.context ?? result;
-    }
+  const matchedPattern = findMatchingPattern(modelName, tokensMap);
+
+  if (matchedPattern) {
+    const result = tokensMap[matchedPattern];
+    return result?.context ?? result;
   }
 
   return undefined;
@@ -181,15 +159,8 @@ function matchModelName(modelName, endpoint = EModelEndpoint.openAI) {
     return modelName;
   }
 
-  const keys = Object.keys(tokensMap);
-  for (let i = keys.length - 1; i >= 0; i--) {
-    const modelKey = keys[i];
-    if (modelName.includes(modelKey)) {
-      return modelKey;
-    }
-  }
-
-  return modelName;
+  const matchedPattern = findMatchingPattern(modelName, tokensMap);
+  return matchedPattern || modelName;
 }
 
 const modelSchema = z.object({
@@ -241,8 +212,47 @@ function processModelData(input) {
   return tokenConfig;
 }
 
+const tiktokenModels = new Set([
+  'text-davinci-003',
+  'text-davinci-002',
+  'text-davinci-001',
+  'text-curie-001',
+  'text-babbage-001',
+  'text-ada-001',
+  'davinci',
+  'curie',
+  'babbage',
+  'ada',
+  'code-davinci-002',
+  'code-davinci-001',
+  'code-cushman-002',
+  'code-cushman-001',
+  'davinci-codex',
+  'cushman-codex',
+  'text-davinci-edit-001',
+  'code-davinci-edit-001',
+  'text-embedding-ada-002',
+  'text-similarity-davinci-001',
+  'text-similarity-curie-001',
+  'text-similarity-babbage-001',
+  'text-similarity-ada-001',
+  'text-search-davinci-doc-001',
+  'text-search-curie-doc-001',
+  'text-search-babbage-doc-001',
+  'text-search-ada-doc-001',
+  'code-search-babbage-code-001',
+  'code-search-ada-code-001',
+  'gpt2',
+  'gpt-4',
+  'gpt-4-0314',
+  'gpt-4-32k',
+  'gpt-4-32k-0314',
+  'gpt-3.5-turbo',
+  'gpt-3.5-turbo-0301',
+]);
+
 module.exports = {
-  tiktokenModels: new Set(models),
+  tiktokenModels,
   maxTokensMap,
   inputSchema,
   modelSchema,

packages/data-provider/package.json

@@ -1,6 +1,6 @@
 {
   "name": "librechat-data-provider",
-  "version": "0.7.2",
+  "version": "0.7.3",
   "description": "data services for librechat apps",
   "main": "dist/index.js",
   "module": "dist/index.es.js",

packages/data-provider/src/config.ts

@@ -12,6 +12,8 @@ export const defaultSocialLogins = ['google', 'facebook', 'openid', 'github', 'd
 export const defaultRetrievalModels = [
   'gpt-4o',
   'gpt-4o-2024-05-13',
+  'gpt-4o-mini',
+  'gpt-4o-mini-2024-07-18',
   'gpt-4-turbo-preview',
   'gpt-3.5-turbo-0125',
   'gpt-4-0125-preview',
@@ -530,7 +532,7 @@ const sharedOpenAIModels = [
 
 export const defaultModels = {
   [EModelEndpoint.azureAssistants]: sharedOpenAIModels,
-  [EModelEndpoint.assistants]: ['gpt-4o', ...sharedOpenAIModels],
+  [EModelEndpoint.assistants]: ['gpt-4o-mini', 'gpt-4o', ...sharedOpenAIModels],
   [EModelEndpoint.google]: [
     'gemini-pro',
     'gemini-pro-vision',
@@ -559,13 +561,12 @@ export const defaultModels = {
     'claude-instant-1-100k',
   ],
   [EModelEndpoint.openAI]: [
+    'gpt-4o-mini',
     'gpt-4o',
     ...sharedOpenAIModels,
     'gpt-4-vision-preview',
     'gpt-3.5-turbo-instruct-0914',
-    'gpt-3.5-turbo-0301',
     'gpt-3.5-turbo-instruct',
-    'text-davinci-003',
   ],
 };
 
@@ -621,6 +622,7 @@ export const supportsBalanceCheck = {
 
 export const visionModels = [
   'gpt-4o',
+  'gpt-4o-mini',
   'gpt-4-turbo',
   'gpt-4-vision',
   'llava',

packages/data-provider/src/schemas.ts

@@ -219,7 +219,7 @@ export enum EAgent {
 
 export const agentOptionSettings = {
   model: {
-    default: 'gpt-4o',
+    default: 'gpt-4o-mini',
   },
   temperature: {
     min: 0,