replace {{current_date}} server side for OpenAI, Anthropic and Assistants

.devcontainer/docker-compose.yml

@@ -1,5 +1,3 @@
-version: "3.8"
-
 services:
   app:
     build:

.env.example

@@ -82,7 +82,7 @@ PROXY=
 #============#
 
 ANTHROPIC_API_KEY=user_provided
-# ANTHROPIC_MODELS=claude-3-5-sonnet-20240620,claude-3-opus-20240229,claude-3-sonnet-20240229,claude-3-haiku-20240307,claude-2.1,claude-2,claude-1.2,claude-1,claude-1-100k,claude-instant-1,claude-instant-1-100k
+# ANTHROPIC_MODELS=claude-3-5-sonnet-20241022,claude-3-5-sonnet-latest,claude-3-5-sonnet-20240620,claude-3-opus-20240229,claude-3-sonnet-20240229,claude-3-haiku-20240307,claude-2.1,claude-2,claude-1.2,claude-1,claude-1-100k,claude-instant-1,claude-instant-1-100k
 # ANTHROPIC_REVERSE_PROXY=
 
 #============#
@@ -146,6 +146,8 @@ GOOGLE_KEY=user_provided
 
 # GOOGLE_TITLE_MODEL=gemini-pro
 
+# GOOGLE_LOC=us-central1
+
 # Google Safety Settings
 # NOTE: These settings apply to both Vertex AI and Gemini API (AI Studio)
 #

Dockerfile

@@ -1,4 +1,4 @@
-# v0.7.5-rc2
+# v0.7.5
 
 # Base node image
 FROM node:20-alpine AS node

Dockerfile.multi

@@ -1,5 +1,5 @@
 # Dockerfile.multi
-# v0.7.5-rc2
+# v0.7.5
 
 # Base for all builds
 FROM node:20-alpine AS base

api/app/clients/AnthropicClient.js

@@ -17,7 +17,7 @@ const {
   parseParamFromPrompt,
   createContextHandlers,
 } = require('./prompts');
-const { getModelMaxTokens, getModelMaxOutputTokens, matchModelName } = require('~/utils');
+const { getModelMaxTokens, getModelMaxOutputTokens, matchModelName, replaceSpecialVars} = require('~/utils');
 const { spendTokens, spendStructuredTokens } = require('~/models/spendTokens');
 const { sleep } = require('~/server/utils');
 const BaseClient = require('./BaseClient');
@@ -494,7 +494,7 @@ class AnthropicClient extends BaseClient {
       identityPrefix = `${identityPrefix}\nYou are ${this.options.modelLabel}`;
     }
 
-    let promptPrefix = (this.options.promptPrefix ?? '').trim();
+    let promptPrefix = replaceSpecialVars((this.options.promptPrefix ?? '').trim());
     if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
       promptPrefix = `${promptPrefix ?? ''}\n${this.options.artifactsPrompt}`.trim();
     }
@@ -682,6 +682,9 @@ class AnthropicClient extends BaseClient {
    */
   checkPromptCacheSupport(modelName) {
     const modelMatch = matchModelName(modelName, EModelEndpoint.anthropic);
+    if (modelMatch.includes('claude-3-5-sonnet-latest')) {
+      return false;
+    }
     if (
       modelMatch === 'claude-3-5-sonnet' ||
       modelMatch === 'claude-3-haiku' ||

api/app/clients/GoogleClient.js

@@ -28,7 +28,7 @@ const {
 } = require('./prompts');
 const BaseClient = require('./BaseClient');
 
-const loc = 'us-central1';
+const loc = process.env.GOOGLE_LOC || 'us-central1';
 const publisher = 'google';
 const endpointPrefix = `https://${loc}-aiplatform.googleapis.com`;
 // const apiEndpoint = loc + '-aiplatform.googleapis.com';
@@ -593,6 +593,8 @@ class GoogleClient extends BaseClient {
 
   createLLM(clientOptions) {
     const model = clientOptions.modelName ?? clientOptions.model;
+    clientOptions.location = loc;
+    clientOptions.endpoint = `${loc}-aiplatform.googleapis.com`;
     if (this.project_id && this.isTextModel) {
       logger.debug('Creating Google VertexAI client');
       return new GoogleVertexAI(clientOptions);

api/app/clients/OllamaClient.js

@@ -60,7 +60,9 @@ class OllamaClient {
     try {
       const ollamaEndpoint = deriveBaseURL(baseURL);
       /** @type {Promise<AxiosResponse<OllamaListResponse>>} */
-      const response = await axios.get(`${ollamaEndpoint}/api/tags`);
+      const response = await axios.get(`${ollamaEndpoint}/api/tags`, {
+        timeout: 5000,
+      });
       models = response.data.models.map((tag) => tag.name);
       return models;
     } catch (error) {

api/app/clients/OpenAIClient.js

@@ -20,6 +20,7 @@ const {
   getModelMaxTokens,
   genAzureChatCompletion,
   getModelMaxOutputTokens,
+  replaceSpecialVars,
 } = require('~/utils');
 const {
   truncateText,
@@ -68,6 +69,8 @@ class OpenAIClient extends BaseClient {
 
     /** @type {OpenAIUsageMetadata | undefined} */
     this.usage;
+    /** @type {boolean|undefined} */
+    this.isO1Model;
   }
 
   // TODO: PluginsClient calls this 3x, unneeded
@@ -98,6 +101,8 @@ class OpenAIClient extends BaseClient {
       this.options.modelOptions,
     );
 
+    this.isO1Model = /\bo1\b/i.test(this.modelOptions.model);
+
     this.defaultVisionModel = this.options.visionModel ?? 'gpt-4-vision-preview';
     if (typeof this.options.attachments?.then === 'function') {
       this.options.attachments.then((attachments) => this.checkVisionRequest(attachments));
@@ -545,9 +550,8 @@ class OpenAIClient extends BaseClient {
       promptPrefix = this.augmentedPrompt + promptPrefix;
     }
 
-    const isO1Model = /\bo1\b/i.test(this.modelOptions.model);
-    if (promptPrefix && !isO1Model) {
-      promptPrefix = `Instructions:\n${promptPrefix.trim()}`;
+    if (promptPrefix && this.isO1Model !== true) {
+      promptPrefix = `Instructions:\n${replaceSpecialVars(promptPrefix.trim())}`;
       instructions = {
         role: 'system',
         name: 'instructions',
@@ -575,7 +579,7 @@ class OpenAIClient extends BaseClient {
     };
 
     /** EXPERIMENTAL */
-    if (promptPrefix && isO1Model) {
+    if (promptPrefix && this.isO1Model === true) {
       const lastUserMessageIndex = payload.findLastIndex((message) => message.role === 'user');
       if (lastUserMessageIndex !== -1) {
         payload[
@@ -839,27 +843,27 @@ class OpenAIClient extends BaseClient {
     }
 
     const titleChatCompletion = async () => {
-      modelOptions.model = model;
+      try {
+        modelOptions.model = model;
 
-      if (this.azure) {
-        modelOptions.model = process.env.AZURE_OPENAI_DEFAULT_MODEL ?? modelOptions.model;
-        this.azureEndpoint = genAzureChatCompletion(this.azure, modelOptions.model, this);
-      }
+        if (this.azure) {
+          modelOptions.model = process.env.AZURE_OPENAI_DEFAULT_MODEL ?? modelOptions.model;
+          this.azureEndpoint = genAzureChatCompletion(this.azure, modelOptions.model, this);
+        }
 
-      const instructionsPayload = [
-        {
-          role: this.options.titleMessageRole ?? (this.isOllama ? 'user' : 'system'),
-          content: `Please generate ${titleInstruction}
+        const instructionsPayload = [
+          {
+            role: this.options.titleMessageRole ?? (this.isOllama ? 'user' : 'system'),
+            content: `Please generate ${titleInstruction}
 
 ${convo}
 
 ||>Title:`,
-        },
-      ];
+          },
+        ];
 
-      const promptTokens = this.getTokenCountForMessage(instructionsPayload[0]);
+        const promptTokens = this.getTokenCountForMessage(instructionsPayload[0]);
 
-      try {
         let useChatCompletion = true;
 
         if (this.options.reverseProxyUrl === CohereConstants.API_URL) {
@@ -1227,6 +1231,11 @@ ${convo}
         opts.defaultHeaders = { ...opts.defaultHeaders, 'api-key': this.apiKey };
       }
 
+      if (this.isO1Model === true && modelOptions.max_tokens != null) {
+        modelOptions.max_completion_tokens = modelOptions.max_tokens;
+        delete modelOptions.max_tokens;
+      }
+
       if (process.env.OPENAI_ORGANIZATION) {
         opts.organization = process.env.OPENAI_ORGANIZATION;
       }
@@ -1301,7 +1310,7 @@ ${convo}
       /** @type {(value: void | PromiseLike<void>) => void} */
       let streamResolve;
 
-      if (modelOptions.stream && /\bo1\b/i.test(modelOptions.model)) {
+      if (modelOptions.stream && this.isO1Model) {
         delete modelOptions.stream;
         delete modelOptions.stop;
       }

api/app/clients/specs/AnthropicClient.test.js

@@ -201,10 +201,10 @@ describe('AnthropicClient', () => {
       );
     });
 
-    it('should add beta header for claude-3-5-sonnet model', () => {
+    it('should add "max-tokens" & "prompt-caching" beta header for claude-3-5-sonnet model', () => {
       const client = new AnthropicClient('test-api-key');
       const modelOptions = {
-        model: 'claude-3-5-sonnet-20240307',
+        model: 'claude-3-5-sonnet-20241022',
       };
       client.setOptions({ modelOptions, promptCache: true });
       const anthropicClient = client.getClient(modelOptions);
@@ -215,7 +215,7 @@ describe('AnthropicClient', () => {
       );
     });
 
-    it('should add beta header for claude-3-haiku model', () => {
+    it('should add "prompt-caching" beta header for claude-3-haiku model', () => {
       const client = new AnthropicClient('test-api-key');
       const modelOptions = {
         model: 'claude-3-haiku-2028',
@@ -229,6 +229,30 @@ describe('AnthropicClient', () => {
       );
     });
 
+    it('should add "prompt-caching" beta header for claude-3-opus model', () => {
+      const client = new AnthropicClient('test-api-key');
+      const modelOptions = {
+        model: 'claude-3-opus-2028',
+      };
+      client.setOptions({ modelOptions, promptCache: true });
+      const anthropicClient = client.getClient(modelOptions);
+      expect(anthropicClient._options.defaultHeaders).toBeDefined();
+      expect(anthropicClient._options.defaultHeaders).toHaveProperty('anthropic-beta');
+      expect(anthropicClient._options.defaultHeaders['anthropic-beta']).toBe(
+        'prompt-caching-2024-07-31',
+      );
+    });
+
+    it('should not add beta header for claude-3-5-sonnet-latest model', () => {
+      const client = new AnthropicClient('test-api-key');
+      const modelOptions = {
+        model: 'anthropic/claude-3-5-sonnet-latest',
+      };
+      client.setOptions({ modelOptions, promptCache: true });
+      const anthropicClient = client.getClient(modelOptions);
+      expect(anthropicClient.defaultHeaders).not.toHaveProperty('anthropic-beta');
+    });
+
     it('should not add beta header for other models', () => {
       const client = new AnthropicClient('test-api-key');
       client.setOptions({

api/models/Preset.js

@@ -3,7 +3,13 @@ const { logger } = require('~/config');
 
 const getPreset = async (user, presetId) => {
   try {
-    return await Preset.findOne({ user, presetId }).lean();
+    // Try to find user-specific preset
+    let preset = await Preset.findOne({ user, presetId }).lean();
+    if (!preset) {
+      // If not found, try to find global preset
+      preset = await Preset.findOne({ presetId, isGlobal: true }).lean();
+    }
+    return preset;
   } catch (error) {
     logger.error('[getPreset] Error getting single preset', error);
     return { message: 'Error getting single preset' };
@@ -15,7 +21,19 @@ module.exports = {
   getPreset,
   getPresets: async (user, filter) => {
     try {
-      const presets = await Preset.find({ ...filter, user }).lean();
+
+      // Fetch user-specific presets
+      const userPresetsPromise = Preset.find({ ...filter, user }).lean();
+      // Fetch global presets
+      const globalPresetsPromise = Preset.find({ ...filter, isGlobal: true }).lean();
+
+      const [userPresets, globalPresets] = await Promise.all([
+        userPresetsPromise,
+        globalPresetsPromise,
+      ]);
+
+      // Combine global and user-specific presets
+      let presets = [...globalPresets, ...userPresets];
       const defaultValue = 10000;
 
       presets.sort((a, b) => {
@@ -28,7 +46,6 @@ module.exports = {
 
         return b.updatedAt - a.updatedAt;
       });
-
       return presets;
     } catch (error) {
       logger.error('[getPresets] Error getting presets', error);
@@ -37,8 +54,32 @@ module.exports = {
   },
   savePreset: async (user, { presetId, newPresetId, defaultPreset, ...preset }) => {
     try {
+      // Check if presetId corresponds to a global preset
+      if (presetId && !newPresetId) {
+        const existingPreset = await Preset.findOne({ presetId, isGlobal: true });
+        if (existingPreset) {
+          // User is attempting to edit a global preset
+          // Create a copy for the user
+          presetId = `user-${user}-${presetId}-${Date.now()}`; // Generate a unique presetId for the user
+
+          const { _id, createdAt, updatedAt, _v,isGlobal, ...presetWithoutIdAndTimestamps } = preset;
+
+          // Prepare the new preset data
+          const newPreset = {
+            presetId: presetId,
+            user: user,
+            isGlobal: false,
+            ...presetWithoutIdAndTimestamps,
+          };
+          // Assign the modified preset back
+          preset = newPreset;
+
+        }
+      }
+
       const setter = { $set: {} };
-      const update = { presetId, ...preset };
+      const { user: _, ...cleanPreset } = preset;
+      const update = { presetId, ...cleanPreset };
       if (preset.tools && Array.isArray(preset.tools)) {
         update.tools =
           preset.tools

api/models/Prompt.js

@@ -7,6 +7,7 @@ const {
   removeGroupFromAllProjects,
 } = require('./Project');
 const { Prompt, PromptGroup } = require('./schema/promptSchema');
+const { escapeRegExp } = require('~/server/utils');
 const { logger } = require('~/config');
 
 /**
@@ -106,7 +107,7 @@ const getAllPromptGroups = async (req, filter) => {
     let searchShared = true;
     let searchSharedOnly = false;
     if (name) {
-      query.name = new RegExp(name, 'i');
+      query.name = new RegExp(escapeRegExp(name), 'i');
     }
     if (!query.category) {
       delete query.category;
@@ -159,7 +160,7 @@ const getPromptGroups = async (req, filter) => {
     let searchShared = true;
     let searchSharedOnly = false;
     if (name) {
-      query.name = new RegExp(name, 'i');
+      query.name = new RegExp(escapeRegExp(name), 'i');
     }
     if (!query.category) {
       delete query.category;

api/models/schema/presetSchema.js

@@ -17,6 +17,11 @@ const presetSchema = mongoose.Schema(
       type: String,
       default: null,
     },
+    isGlobal: {
+      type: Boolean,
+      default: false,
+      index: true,
+    },
     defaultPreset: {
       type: Boolean,
     },

api/models/tx.js

@@ -40,9 +40,9 @@ const tokenValues = Object.assign(
     'o1-preview': { prompt: 15, completion: 60 },
     'o1-mini': { prompt: 3, completion: 12 },
     o1: { prompt: 15, completion: 60 },
-    'gpt-4o-2024-08-06': { prompt: 2.5, completion: 10 },
     'gpt-4o-mini': { prompt: 0.15, completion: 0.6 },
-    'gpt-4o': { prompt: 5, completion: 15 },
+    'gpt-4o': { prompt: 2.5, completion: 10 },
+    'gpt-4o-2024-05-13': { prompt: 5, completion: 15 },
     'gpt-4-1106': { prompt: 10, completion: 30 },
     'gpt-3.5-turbo-0125': { prompt: 0.5, completion: 1.5 },
     'claude-3-opus': { prompt: 15, completion: 75 },
@@ -104,8 +104,8 @@ const getValueKey = (model, endpoint) => {
     return 'o1-mini';
   } else if (modelName.includes('o1')) {
     return 'o1';
-  } else if (modelName.includes('gpt-4o-2024-08-06')) {
-    return 'gpt-4o-2024-08-06';
+  } else if (modelName.includes('gpt-4o-2024-05-13')) {
+    return 'gpt-4o-2024-05-13';
   } else if (modelName.includes('gpt-4o-mini')) {
     return 'gpt-4o-mini';
   } else if (modelName.includes('gpt-4o')) {