From 2e661689604bc8175ea10be637494712dc6982f1 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Fri, 3 Apr 2026 10:05:10 +0000
Subject: [PATCH 1/3] Initial plan


From 873118293959013c625dd5915012b92c79f7b46e Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Fri, 3 Apr 2026 10:22:03 +0000
Subject: [PATCH 2/3] feat: Add Pollinations API integration with nova-fast as
 default model
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Add POLLINATIONS_API_KEY / POLLINATIONS_BASE_URL to .env.example
- Replace active custom endpoint examples in librechat.example.yaml with
  Pollinations (nova-fast default); old examples commented out
- packages/api/src/pollinations/classify.ts: intent classification +
  image-prompt enhancement via nova-fast
- packages/api/src/pollinations/service.ts: orchestration layer —
  text → nova-fast, image → enhance+zimage, video → placeholder
- packages/api/src/pollinations/index.ts: public exports
- packages/api/src/pollinations/classify.spec.ts: Jest tests for all
  intent paths, error cases, and multi-turn conversation
- packages/api/src/index.ts: export Pollinations module
- README.md: Pollinations integration section with setup instructions

Agent-Logs-Url: https://github.com/verdjs/LibreChat/sessions/ace58aa9-0e14-4a38-9814-baa691114088

Co-authored-by: verdjs <225778361+verdjs@users.noreply.github.com>
---
 .env.example                                  |   9 +
 README.md                                     |  28 ++
 librechat.example.yaml                        | 178 ++++++-------
 packages/api/src/index.ts                     |   2 +
 .../api/src/pollinations/classify.spec.ts     | 244 ++++++++++++++++++
 packages/api/src/pollinations/classify.ts     |  81 ++++++
 packages/api/src/pollinations/index.ts        |  11 +
 packages/api/src/pollinations/service.ts      | 128 +++++++++
 8 files changed, 585 insertions(+), 96 deletions(-)
 create mode 100644 packages/api/src/pollinations/classify.spec.ts
 create mode 100644 packages/api/src/pollinations/classify.ts
 create mode 100644 packages/api/src/pollinations/index.ts
 create mode 100644 packages/api/src/pollinations/service.ts

diff --git a/.env.example b/.env.example
index db09bb471f7..cfcca874f3c 100644
--- a/.env.example
+++ b/.env.example
@@ -251,6 +251,15 @@ GOOGLE_KEY=user_provided
 # Vertex AI model for image generation (defaults to gemini-2.5-flash-image)
 # GEMINI_IMAGE_MODEL=gemini-2.5-flash-image
 
+#===================#
+# Pollinations API  #
+#===================#
+
+# Pollinations is the primary generation backend for text, image, and video.
+# Get your API key at: https://enter.pollinations.ai
+POLLINATIONS_API_KEY=your_pollinations_api_key
+# POLLINATIONS_BASE_URL=https://gen.pollinations.ai
+
 #============#
 # OpenAI     #
 #============#
diff --git a/README.md b/README.md
index a7f68d9a920..254cdb33446 100644
--- a/README.md
+++ b/README.md
@@ -155,6 +155,34 @@ Open source, actively developed, and built for anyone who values control over th
 
 ---
 
+## 🌸 Pollinations Integration
+
+LibreChat ships with **[Pollinations](https://pollinations.ai)** as its primary generation backend.
+Every message is automatically routed through a lightweight intent classifier before reaching the model:
+
+| Intent | Flow |
+|--------|------|
+| **Text** | Classified by `nova-fast` → answered by `nova-fast` (or any model you choose) |
+| **Image** | Classified by `nova-fast` → prompt enhanced by `nova-fast` → image generated with `zimage` |
+| **Video** | Classified by `nova-fast` → _coming soon_ placeholder returned |
+
+### Setup
+
+1. Get a free API key at <https://enter.pollinations.ai>.
+2. Add it to your `.env`:
+   ```
+   POLLINATIONS_API_KEY=your_pollinations_api_key
+   ```
+3. Copy `librechat.example.yaml` → `librechat.yaml` (or use `CONFIG_PATH`). The **Pollinations** endpoint is already the first entry under `endpoints.custom`.
+
+### Available models (sample)
+
+`nova-fast` · `openai` · `openai-fast` · `openai-large` · `gemini-fast` · `claude-fast` · `deepseek` · `mistral` · `qwen-coder` · `grok` · `perplexity-fast` · `kimi` — plus many more. Set `fetch: true` in the YAML to auto-populate all current models from the Pollinations `/v1/models` endpoint.
+
+> **Key:** Use an `sk_` prefixed secret key for server-side requests. `pk_` keys work client-side but are rate-limited.
+
+---
+
 ## 🌐 Resources
 
 **GitHub Repo:**
diff --git a/librechat.example.yaml b/librechat.example.yaml
index 03bb5f5bc25..904a0e1967a 100644
--- a/librechat.example.yaml
+++ b/librechat.example.yaml
@@ -336,114 +336,100 @@ endpoints:
   #     #     deploymentName: claude-3-5-haiku@20241022  # Override for this model
 
   custom:
-    # Groq Example
-    - name: 'groq'
-      apiKey: '${GROQ_API_KEY}'
-      baseURL: 'https://api.groq.com/openai/v1/'
+    # Pollinations — primary generation backend (text, image, video)
+    # Get your API key at: https://enter.pollinations.ai
+    # Docs: https://gen.pollinations.ai/api/docs
+    - name: 'Pollinations'
+      apiKey: '${POLLINATIONS_API_KEY}'
+      baseURL: 'https://gen.pollinations.ai'
       models:
         default:
-          - 'llama3-70b-8192'
-          - 'llama3-8b-8192'
-          - 'llama2-70b-4096'
-          - 'mixtral-8x7b-32768'
-          - 'gemma-7b-it'
-        fetch: false
+          - 'nova-fast'
+          - 'openai'
+          - 'openai-fast'
+          - 'openai-large'
+          - 'claude-fast'
+          - 'claude'
+          - 'gemini'
+          - 'gemini-fast'
+          - 'deepseek'
+          - 'mistral'
+          - 'qwen-coder'
+          - 'nova'
+          - 'grok'
+          - 'perplexity-fast'
+          - 'kimi'
+        fetch: true
       titleConvo: true
-      titleModel: 'mixtral-8x7b-32768'
-      modelDisplayLabel: 'groq'
-
-    # Mistral AI Example
-    - name: 'Mistral' # Unique name for the endpoint
-      # For `apiKey` and `baseURL`, you can use environment variables that you define.
-      # recommended environment variables:
-      apiKey: '${MISTRAL_API_KEY}'
-      baseURL: 'https://api.mistral.ai/v1'
-
-      # Models configuration
-      models:
-        # List of default models to use. At least one value is required.
-        default: ['mistral-tiny', 'mistral-small', 'mistral-medium']
-        # Fetch option: Set to true to fetch models from API.
-        fetch: true # Defaults to false.
-
-      # Optional configurations
-
-      # Title Conversation setting
-      titleConvo: true # Set to true to enable title conversation
-
-      # Title Method: Choose between "completion" or "functions".
-      # titleMethod: "completion"  # Defaults to "completion" if omitted.
-
-      # Title Model: Specify the model to use for titles.
-      titleModel: 'mistral-tiny' # Defaults to "gpt-3.5-turbo" if omitted.
-
-      # Summarize setting: Set to true to enable summarization.
-      # summarize: false
-
-      # Summary Model: Specify the model to use if summarization is enabled.
-      # summaryModel: "mistral-tiny"  # Defaults to "gpt-3.5-turbo" if omitted.
+      titleModel: 'nova-fast'
+      modelDisplayLabel: 'Pollinations'
+      iconURL: 'https://pollinations.ai/favicon.ico'
 
-      # The label displayed for the AI model in messages.
-      modelDisplayLabel: 'Mistral' # Default is "AI" when not set.
+    # ---------------------------------------------------------------
+    # The following custom endpoint examples are disabled by default.
+    # Pollinations (above) is the primary configured backend.
+    # Uncomment any section below if you need that specific provider.
+    # ---------------------------------------------------------------
 
-      # Add additional parameters to the request. Default params will be overwritten.
-      # addParams:
-      # safe_prompt: true # This field is specific to Mistral AI: https://docs.mistral.ai/api/
+    # Groq Example
+    # - name: 'groq'
+    #   apiKey: '${GROQ_API_KEY}'
+    #   baseURL: 'https://api.groq.com/openai/v1/'
+    #   models:
+    #     default: ['llama3-70b-8192', 'llama3-8b-8192', 'mixtral-8x7b-32768']
+    #     fetch: false
+    #   titleConvo: true
+    #   titleModel: 'mixtral-8x7b-32768'
+    #   modelDisplayLabel: 'groq'
 
-      # Drop Default params parameters from the request. See default params in guide linked below.
-      # NOTE: For Mistral, it is necessary to drop the following parameters or you will encounter a 422 Error:
-      dropParams: ['stop', 'user', 'frequency_penalty', 'presence_penalty']
+    # Mistral AI Example
+    # - name: 'Mistral'
+    #   apiKey: '${MISTRAL_API_KEY}'
+    #   baseURL: 'https://api.mistral.ai/v1'
+    #   models:
+    #     default: ['mistral-tiny', 'mistral-small', 'mistral-medium']
+    #     fetch: true
+    #   titleConvo: true
+    #   titleModel: 'mistral-tiny'
+    #   modelDisplayLabel: 'Mistral'
+    #   dropParams: ['stop', 'user', 'frequency_penalty', 'presence_penalty']
 
     # OpenRouter Example
-    - name: 'OpenRouter'
-      # For `apiKey` and `baseURL`, you can use environment variables that you define.
-      # recommended environment variables:
-      apiKey: '${OPENROUTER_KEY}'
-      baseURL: 'https://openrouter.ai/api/v1'
-      headers:
-        x-librechat-body-parentmessageid: '{{LIBRECHAT_BODY_PARENTMESSAGEID}}'
-      models:
-        default: ['meta-llama/llama-3-70b-instruct']
-        fetch: true
-      titleConvo: true
-      titleModel: 'meta-llama/llama-3-70b-instruct'
-      # Recommended: Drop the stop parameter from the request as Openrouter models use a variety of stop tokens.
-      dropParams: ['stop']
-      modelDisplayLabel: 'OpenRouter'
+    # - name: 'OpenRouter'
+    #   apiKey: '${OPENROUTER_KEY}'
+    #   baseURL: 'https://openrouter.ai/api/v1'
+    #   models:
+    #     default: ['meta-llama/llama-3-70b-instruct']
+    #     fetch: true
+    #   titleConvo: true
+    #   titleModel: 'meta-llama/llama-3-70b-instruct'
+    #   dropParams: ['stop']
+    #   modelDisplayLabel: 'OpenRouter'
 
     # Helicone Example
-    - name: 'Helicone'
-      # For `apiKey` and `baseURL`, you can use environment variables that you define.
-      # recommended environment variables:
-      apiKey: '${HELICONE_KEY}'
-      baseURL: 'https://ai-gateway.helicone.ai'
-      headers:
-        x-librechat-body-parentmessageid: '{{LIBRECHAT_BODY_PARENTMESSAGEID}}'
-      models:
-        default:
-          ['gpt-4o-mini', 'claude-4.5-sonnet', 'llama-3.1-8b-instruct', 'gemini-2.5-flash-lite']
-        fetch: true
-      titleConvo: true
-      titleModel: 'gpt-4o-mini'
-      modelDisplayLabel: 'Helicone'
-      iconURL: https://marketing-assets-helicone.s3.us-west-2.amazonaws.com/helicone.png
+    # - name: 'Helicone'
+    #   apiKey: '${HELICONE_KEY}'
+    #   baseURL: 'https://ai-gateway.helicone.ai'
+    #   models:
+    #     default: ['gpt-4o-mini', 'gemini-2.5-flash-lite']
+    #     fetch: true
+    #   titleConvo: true
+    #   titleModel: 'gpt-4o-mini'
+    #   modelDisplayLabel: 'Helicone'
 
     # Portkey AI Example
-    - name: 'Portkey'
-      apiKey: 'dummy'
-      baseURL: 'https://api.portkey.ai/v1'
-      headers:
-        x-portkey-api-key: '${PORTKEY_API_KEY}'
-        x-portkey-virtual-key: '${PORTKEY_OPENAI_VIRTUAL_KEY}'
-      models:
-        default: ['gpt-4o-mini', 'gpt-4o', 'chatgpt-4o-latest']
-        fetch: true
-      titleConvo: true
-      titleModel: 'current_model'
-      summarize: false
-      summaryModel: 'current_model'
-      modelDisplayLabel: 'Portkey'
-      iconURL: https://images.crunchbase.com/image/upload/c_pad,f_auto,q_auto:eco,dpr_1/rjqy7ghvjoiu4cd1xjbf
+    # - name: 'Portkey'
+    #   apiKey: 'dummy'
+    #   baseURL: 'https://api.portkey.ai/v1'
+    #   headers:
+    #     x-portkey-api-key: '${PORTKEY_API_KEY}'
+    #     x-portkey-virtual-key: '${PORTKEY_OPENAI_VIRTUAL_KEY}'
+    #   models:
+    #     default: ['gpt-4o-mini', 'gpt-4o']
+    #     fetch: true
+    #   titleConvo: true
+    #   titleModel: 'current_model'
+    #   modelDisplayLabel: 'Portkey'
 
   # AWS Bedrock Example
   # Note: Bedrock endpoint is configured via environment variables
diff --git a/packages/api/src/index.ts b/packages/api/src/index.ts
index d4b6ac95422..c8766a47aba 100644
--- a/packages/api/src/index.ts
+++ b/packages/api/src/index.ts
@@ -39,6 +39,8 @@ export * from './agents';
 export * from './prompts';
 /* Endpoints */
 export * from './endpoints';
+/* Pollinations */
+export * from './pollinations';
 /* Files */
 export * from './files';
 /* Storage */
diff --git a/packages/api/src/pollinations/classify.spec.ts b/packages/api/src/pollinations/classify.spec.ts
new file mode 100644
index 00000000000..da710cb68b3
--- /dev/null
+++ b/packages/api/src/pollinations/classify.spec.ts
@@ -0,0 +1,244 @@
+import axios from 'axios';
+
+jest.mock('axios');
+
+import { classifyIntent, enhanceImagePrompt } from './classify';
+import { processPollinationsRequest, DEFAULT_TEXT_MODEL, IMAGE_MODEL } from './service';
+
+const mockedAxios = axios as jest.Mocked<typeof axios>;
+
+const TEST_API_KEY = 'sk-test-key';
+const TEST_MESSAGE = 'Hello, tell me a joke';
+const TEST_IMAGE_MESSAGE = 'Draw a cat in space';
+const TEST_VIDEO_MESSAGE = 'Make a video of a sunset';
+
+function makeChatResponse(content: string) {
+  return { data: { choices: [{ message: { content } }] } };
+}
+
+function makeImageResponse(url: string) {
+  return { data: { data: [{ url }] } };
+}
+
+function setupAxiosClientMock(): jest.Mock {
+  const mockClientPost = jest.fn();
+  mockedAxios.create.mockReturnValue({ post: mockClientPost } as ReturnType<typeof axios.create>);
+  return mockClientPost;
+}
+
+// ─── classifyIntent ───────────────────────────────────────────────────────────
+
+describe('classifyIntent', () => {
+  let mockClientPost: jest.Mock;
+
+  beforeEach(() => {
+    jest.clearAllMocks();
+    mockClientPost = setupAxiosClientMock();
+  });
+
+  it('returns "text" for a general text message', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('text'));
+    const result = await classifyIntent(TEST_MESSAGE, TEST_API_KEY);
+    expect(result).toBe('text');
+  });
+
+  it('returns "image" when model responds with "image"', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('image'));
+    const result = await classifyIntent(TEST_IMAGE_MESSAGE, TEST_API_KEY);
+    expect(result).toBe('image');
+  });
+
+  it('returns "video" when model responds with "video"', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('video'));
+    const result = await classifyIntent(TEST_VIDEO_MESSAGE, TEST_API_KEY);
+    expect(result).toBe('video');
+  });
+
+  it('defaults to "text" for unrecognised model output', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('unknown-intent'));
+    const result = await classifyIntent(TEST_MESSAGE, TEST_API_KEY);
+    expect(result).toBe('text');
+  });
+
+  it('calls Pollinations /v1/chat/completions with nova-fast model', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('text'));
+    await classifyIntent(TEST_MESSAGE, TEST_API_KEY);
+
+    expect(mockClientPost).toHaveBeenCalledWith(
+      '/v1/chat/completions',
+      expect.objectContaining({ model: 'nova-fast' }),
+    );
+  });
+
+  it('trims and lowercases the model response', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('  IMAGE  '));
+    const result = await classifyIntent(TEST_IMAGE_MESSAGE, TEST_API_KEY);
+    expect(result).toBe('image');
+  });
+});
+
+// ─── enhanceImagePrompt ───────────────────────────────────────────────────────
+
+describe('enhanceImagePrompt', () => {
+  let mockClientPost: jest.Mock;
+
+  beforeEach(() => {
+    jest.clearAllMocks();
+    mockClientPost = setupAxiosClientMock();
+  });
+
+  it('returns the enhanced prompt from the model', async () => {
+    const enhanced = 'A majestic cat floating in a vivid, star-filled outer space background';
+    mockClientPost.mockResolvedValueOnce(makeChatResponse(enhanced));
+    const result = await enhanceImagePrompt(TEST_IMAGE_MESSAGE, TEST_API_KEY);
+    expect(result).toBe(enhanced);
+  });
+
+  it('falls back to the original message when model returns empty content', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse(''));
+    const result = await enhanceImagePrompt(TEST_IMAGE_MESSAGE, TEST_API_KEY);
+    expect(result).toBe(TEST_IMAGE_MESSAGE);
+  });
+
+  it('calls Pollinations /v1/chat/completions with nova-fast model', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('enhanced prompt'));
+    await enhanceImagePrompt(TEST_IMAGE_MESSAGE, TEST_API_KEY);
+
+    expect(mockClientPost).toHaveBeenCalledWith(
+      '/v1/chat/completions',
+      expect.objectContaining({ model: 'nova-fast' }),
+    );
+  });
+});
+
+// ─── processPollinationsRequest ───────────────────────────────────────────────
+
+describe('processPollinationsRequest', () => {
+  let mockClientPost: jest.Mock;
+
+  beforeEach(() => {
+    jest.clearAllMocks();
+    mockClientPost = setupAxiosClientMock();
+  });
+
+  it('routes text intent to nova-fast and returns text result', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('text'));
+    mockedAxios.post.mockResolvedValueOnce({
+      data: { choices: [{ message: { content: 'A joke!' } }] },
+    });
+
+    const result = await processPollinationsRequest({
+      userMessage: TEST_MESSAGE,
+      apiKey: TEST_API_KEY,
+    });
+
+    expect(result.type).toBe('text');
+    if (result.type === 'text') {
+      expect(result.content).toBe('A joke!');
+      expect(result.model).toBe(DEFAULT_TEXT_MODEL); // 'nova-fast'
+    }
+  });
+
+  it('respects caller-supplied model override for text intent', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('text'));
+    mockedAxios.post.mockResolvedValueOnce({
+      data: { choices: [{ message: { content: 'Hi' } }] },
+    });
+
+    const result = await processPollinationsRequest({
+      userMessage: TEST_MESSAGE,
+      apiKey: TEST_API_KEY,
+      model: 'openai-large',
+    });
+
+    expect(result.type).toBe('text');
+    if (result.type === 'text') {
+      expect(result.model).toBe('openai-large');
+    }
+  });
+
+  it('routes image intent: enhances prompt then calls /v1/images/generations with zimage', async () => {
+    const enhanced = 'A detailed cosmic cat';
+    mockClientPost
+      .mockResolvedValueOnce(makeChatResponse('image')) // classify
+      .mockResolvedValueOnce(makeChatResponse(enhanced)); // enhance
+    mockedAxios.post.mockResolvedValueOnce(makeImageResponse('https://cdn.pollinations.ai/cat.jpg'));
+
+    const result = await processPollinationsRequest({
+      userMessage: TEST_IMAGE_MESSAGE,
+      apiKey: TEST_API_KEY,
+    });
+
+    expect(result.type).toBe('image');
+    if (result.type === 'image') {
+      expect(result.enhancedPrompt).toBe(enhanced);
+      expect(result.url).toBe('https://cdn.pollinations.ai/cat.jpg');
+    }
+
+    expect(mockedAxios.post).toHaveBeenCalledWith(
+      expect.stringContaining('/v1/images/generations'),
+      expect.objectContaining({ model: IMAGE_MODEL, prompt: enhanced }),
+      expect.any(Object),
+    );
+  });
+
+  it('throws when image generation returns no URL', async () => {
+    const enhanced = 'A detailed cosmic cat';
+    mockClientPost
+      .mockResolvedValueOnce(makeChatResponse('image'))
+      .mockResolvedValueOnce(makeChatResponse(enhanced));
+    mockedAxios.post.mockResolvedValueOnce({ data: { data: [] } });
+
+    await expect(
+      processPollinationsRequest({ userMessage: TEST_IMAGE_MESSAGE, apiKey: TEST_API_KEY }),
+    ).rejects.toThrow('no URL');
+  });
+
+  it('throws when text completion returns no content', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('text'));
+    mockedAxios.post.mockResolvedValueOnce({ data: { choices: [] } });
+
+    await expect(
+      processPollinationsRequest({ userMessage: TEST_MESSAGE, apiKey: TEST_API_KEY }),
+    ).rejects.toThrow('no content');
+  });
+
+  it('returns "coming soon" placeholder for video intent', async () => {
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('video'));
+
+    const result = await processPollinationsRequest({
+      userMessage: TEST_VIDEO_MESSAGE,
+      apiKey: TEST_API_KEY,
+    });
+
+    expect(result.type).toBe('video');
+    if (result.type === 'video') {
+      expect(result.message).toMatch(/coming soon/i);
+    }
+  });
+
+  it('passes the full conversation history for multi-turn text requests', async () => {
+    const history = [
+      { role: 'user', content: 'Hello' },
+      { role: 'assistant', content: 'Hi there!' },
+      { role: 'user', content: TEST_MESSAGE },
+    ];
+
+    mockClientPost.mockResolvedValueOnce(makeChatResponse('text'));
+    mockedAxios.post.mockResolvedValueOnce({
+      data: { choices: [{ message: { content: 'reply' } }] },
+    });
+
+    await processPollinationsRequest({
+      userMessage: TEST_MESSAGE,
+      apiKey: TEST_API_KEY,
+      messages: history,
+    });
+
+    expect(mockedAxios.post).toHaveBeenCalledWith(
+      expect.stringContaining('/v1/chat/completions'),
+      expect.objectContaining({ messages: history }),
+      expect.any(Object),
+    );
+  });
+});
diff --git a/packages/api/src/pollinations/classify.ts b/packages/api/src/pollinations/classify.ts
new file mode 100644
index 00000000000..f7980dc40d2
--- /dev/null
+++ b/packages/api/src/pollinations/classify.ts
@@ -0,0 +1,81 @@
+import axios from 'axios';
+
+const DEFAULT_BASE_URL = 'https://gen.pollinations.ai';
+
+/** Classifier model — Amazon Nova Micro via Pollinations */
+const CLASSIFIER_MODEL = 'nova-fast';
+
+/** System prompt for intent classification */
+const CLASSIFY_SYSTEM =
+  'You are an intent classifier. Given a user message, respond with exactly one lowercase word:\n' +
+  '- "image" if the user wants to generate an image, picture, photo, or illustration\n' +
+  '- "video" if the user wants to generate a video, animation, or clip\n' +
+  '- "text" for all other requests (questions, coding, analysis, conversation, etc.)';
+
+/** System prompt for image prompt enhancement */
+const ENHANCE_SYSTEM =
+  'You are an image prompt enhancer. Given a user image request, rewrite it as a detailed, ' +
+  'high-quality image generation prompt. Be concise but descriptive. Return only the enhanced ' +
+  'prompt with no explanation.';
+
+export type IntentType = 'text' | 'image' | 'video';
+
+interface ChatCompletionResponse {
+  choices: Array<{
+    message: {
+      content: string;
+    };
+  }>;
+}
+
+function buildClient(baseURL: string, apiKey: string) {
+  return axios.create({
+    baseURL,
+    headers: {
+      Authorization: `Bearer ${apiKey}`,
+      'Content-Type': 'application/json',
+    },
+  });
+}
+
+/**
+ * Classifies a user message's intent as 'text', 'image', or 'video'
+ * using Pollinations `nova-fast` model.
+ */
+export async function classifyIntent(userMessage: string, apiKey: string): Promise<IntentType> {
+  const baseURL = process.env.POLLINATIONS_BASE_URL ?? DEFAULT_BASE_URL;
+  const client = buildClient(baseURL, apiKey);
+
+  const response = await client.post<ChatCompletionResponse>('/v1/chat/completions', {
+    model: CLASSIFIER_MODEL,
+    messages: [
+      { role: 'system', content: CLASSIFY_SYSTEM },
+      { role: 'user', content: userMessage },
+    ],
+    temperature: 0,
+  });
+
+  const raw = response.data.choices[0]?.message?.content?.trim().toLowerCase() ?? 'text';
+  if (raw === 'image' || raw === 'video') return raw;
+  return 'text';
+}
+
+/**
+ * Enhances a user's image description into a detailed image generation prompt
+ * using Pollinations `nova-fast` model.
+ */
+export async function enhanceImagePrompt(userMessage: string, apiKey: string): Promise<string> {
+  const baseURL = process.env.POLLINATIONS_BASE_URL ?? DEFAULT_BASE_URL;
+  const client = buildClient(baseURL, apiKey);
+
+  const response = await client.post<ChatCompletionResponse>('/v1/chat/completions', {
+    model: CLASSIFIER_MODEL,
+    messages: [
+      { role: 'system', content: ENHANCE_SYSTEM },
+      { role: 'user', content: userMessage },
+    ],
+    temperature: 0.7,
+  });
+
+  return response.data.choices[0]?.message?.content?.trim() ?? userMessage;
+}
diff --git a/packages/api/src/pollinations/index.ts b/packages/api/src/pollinations/index.ts
new file mode 100644
index 00000000000..efd878e3361
--- /dev/null
+++ b/packages/api/src/pollinations/index.ts
@@ -0,0 +1,11 @@
+export { classifyIntent, enhanceImagePrompt } from './classify';
+export { processPollinationsRequest, DEFAULT_TEXT_MODEL, IMAGE_MODEL } from './service';
+export type { IntentType } from './classify';
+export type {
+  PollinationsRequest,
+  PollinationsResult,
+  PollinationsTextResult,
+  PollinationsImageResult,
+  PollinationsVideoResult,
+  PollinationsMessage,
+} from './service';
diff --git a/packages/api/src/pollinations/service.ts b/packages/api/src/pollinations/service.ts
new file mode 100644
index 00000000000..3e4f4c326ec
--- /dev/null
+++ b/packages/api/src/pollinations/service.ts
@@ -0,0 +1,128 @@
+import axios from 'axios';
+import { classifyIntent, enhanceImagePrompt } from './classify';
+import type { IntentType } from './classify';
+
+const DEFAULT_BASE_URL = 'https://gen.pollinations.ai';
+
+/** Default text model — also used for intent classification */
+export const DEFAULT_TEXT_MODEL = 'nova-fast';
+/** Image generation model */
+export const IMAGE_MODEL = 'zimage';
+
+export interface PollinationsTextResult {
+  type: 'text';
+  content: string;
+  model: string;
+}
+
+export interface PollinationsImageResult {
+  type: 'image';
+  url: string;
+  enhancedPrompt: string;
+}
+
+export interface PollinationsVideoResult {
+  type: 'video';
+  message: string;
+}
+
+export type PollinationsResult =
+  | PollinationsTextResult
+  | PollinationsImageResult
+  | PollinationsVideoResult;
+
+export interface PollinationsMessage {
+  role: string;
+  content: string;
+}
+
+export interface PollinationsRequest {
+  /** The user's most recent message */
+  userMessage: string;
+  /** Pollinations API key (`sk_` prefix for server-side use) */
+  apiKey: string;
+  /** Override the text model (defaults to `nova-fast`) */
+  model?: string;
+  /** Full conversation history for multi-turn text requests */
+  messages?: PollinationsMessage[];
+}
+
+interface ImageGenerationResponse {
+  data: Array<{ url?: string; b64_json?: string }>;
+}
+
+/**
+ * Processes a Pollinations request by:
+ * 1. Calling `nova-fast` to classify the user's intent (text / image / video).
+ * 2. Routing to the appropriate generation flow:
+ *    - **image** → enhance prompt with `nova-fast`, generate with `zimage`
+ *    - **text**  → chat completion with `nova-fast` (or caller-supplied model)
+ *    - **video** → placeholder "coming soon" response
+ */
+export async function processPollinationsRequest({
+  userMessage,
+  apiKey,
+  model,
+  messages = [],
+}: PollinationsRequest): Promise<PollinationsResult> {
+  const baseURL = process.env.POLLINATIONS_BASE_URL ?? DEFAULT_BASE_URL;
+  const intent: IntentType = await classifyIntent(userMessage, apiKey);
+
+  if (intent === 'video') {
+    return {
+      type: 'video',
+      message: 'Video generation is coming soon! Stay tuned.',
+    };
+  }
+
+  if (intent === 'image') {
+    const enhancedPrompt = await enhanceImagePrompt(userMessage, apiKey);
+
+    const response = await axios.post<ImageGenerationResponse>(
+      `${baseURL}/v1/images/generations`,
+      {
+        model: IMAGE_MODEL,
+        prompt: enhancedPrompt,
+        size: '1024x1024',
+        response_format: 'url',
+        n: 1,
+      },
+      {
+        headers: {
+          Authorization: `Bearer ${apiKey}`,
+          'Content-Type': 'application/json',
+        },
+      },
+    );
+
+    const url = response.data.data[0]?.url;
+    if (!url) {
+      throw new Error('Pollinations image generation returned no URL.');
+    }
+
+    return { type: 'image', url, enhancedPrompt };
+  }
+
+  // text
+  const textModel = model ?? DEFAULT_TEXT_MODEL;
+  const allMessages: PollinationsMessage[] =
+    messages.length > 0 ? messages : [{ role: 'user', content: userMessage }];
+
+  const response = await axios.post<{ choices: Array<{ message: { content: string } }> }>(
+    `${baseURL}/v1/chat/completions`,
+    { model: textModel, messages: allMessages },
+    {
+      headers: {
+        Authorization: `Bearer ${apiKey}`,
+        'Content-Type': 'application/json',
+      },
+    },
+  );
+
+  const content = response.data.choices[0]?.message?.content;
+  if (content == null) {
+    throw new Error('Pollinations text completion returned no content.');
+  }
+
+  return { type: 'text', content, model: textModel };
+}

From 0710eeadee2b023598d1f04c603a6b1ff62ddc57 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Fri, 3 Apr 2026 10:36:59 +0000
Subject: [PATCH 3/3] =?UTF-8?q?feat:=20Add=20GUEST=5FMODE=20=E2=80=94=20by?=
 =?UTF-8?q?pass=20login,=20auto-authenticate=20as=20persistent=20guest=20u?=
 =?UTF-8?q?ser?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

When GUEST_MODE=true:
- POST /api/auth/refresh creates/reuses a shared guest user and returns a
  real JWT even without a refresh token cookie
- Frontend suppresses all login redirects (AuthContext, useAuthRedirect,
  Root, ChatRoute, Startup all guard with guestMode flag)
- Startup.tsx redirects /login → /c/new immediately when guestMode
- Guest account uses a random non-bcrypt password so it cannot be used
  for local password auth
- TStartupConfig extended with guestMode?: boolean
- .env.example documents GUEST_MODE=false

Agent-Logs-Url: https://github.com/verdjs/LibreChat/sessions/d5aba8f2-9ba1-41a7-b47d-7c07e46a2acc

Co-authored-by: verdjs <225778361+verdjs@users.noreply.github.com>
---
 .env.example                             |  6 ++++
 api/server/controllers/AuthController.js | 45 ++++++++++++++++++++++++
 api/server/routes/config.js              |  2 ++
 client/src/hooks/AuthContext.tsx         | 10 ++++--
 client/src/routes/ChatRoute.tsx          |  2 +-
 client/src/routes/Layouts/Startup.tsx    |  6 +++-
 client/src/routes/Root.tsx               |  2 +-
 client/src/routes/useAuthRedirect.ts     |  8 ++++-
 packages/data-provider/src/config.ts     |  1 +
 9 files changed, 75 insertions(+), 7 deletions(-)

diff --git a/.env.example b/.env.example
index cfcca874f3c..74487079725 100644
--- a/.env.example
+++ b/.env.example
@@ -463,6 +463,12 @@ ILLEGAL_MODEL_REQ_SCORE=5
 # Registration and Login #
 #========================#
 
+# Set GUEST_MODE=true to bypass all login requirements.
+# Visitors are automatically authenticated as a shared guest user — no account needed.
+# The guest user is created on first launch and reused on every subsequent request.
+# Everything else (chat, conversations, settings) works as normal.
+GUEST_MODE=false
+
 ALLOW_EMAIL_LOGIN=true
 ALLOW_REGISTRATION=true
 ALLOW_SOCIAL_LOGIN=false
diff --git a/api/server/controllers/AuthController.js b/api/server/controllers/AuthController.js
index eb44feffa4c..b2455ea0ec3 100644
--- a/api/server/controllers/AuthController.js
+++ b/api/server/controllers/AuthController.js
@@ -1,8 +1,10 @@
 const cookies = require('cookie');
 const jwt = require('jsonwebtoken');
 const openIdClient = require('openid-client');
+const { randomBytes } = require('node:crypto');
 const { logger } = require('@librechat/data-schemas');
 const { isEnabled, findOpenIDUser } = require('@librechat/api');
+const { SystemRoles } = require('librechat-data-provider');
 const {
   requestPasswordReset,
   setOpenIDAuthTokens,
@@ -13,6 +15,7 @@ const {
 const {
   deleteAllUserSessions,
   getUserById,
+  createUser,
   findSession,
   updateUser,
   findUser,
@@ -20,6 +23,39 @@ const {
 const { getGraphApiToken } = require('~/server/services/GraphTokenService');
 const { getOpenIdConfig, getOpenIdEmail } = require('~/strategies');
 
+const GUEST_EMAIL = process.env.GUEST_USER_EMAIL || 'guest@librechat.local';
+
+/**
+ * Finds or creates the shared guest user used in GUEST_MODE.
+ * The guest user is a regular local user with a stable email address and no password.
+ * It is created once and reused for every unauthenticated request.
+ * @returns {Promise<object>} User document without sensitive fields
+ */
+async function getOrCreateGuestUser() {
+  let user = await findUser({ email: GUEST_EMAIL }, '-password -__v -totpSecret -backupCodes');
+  if (!user) {
+    const created = await createUser(
+      {
+        provider: 'local',
+        email: GUEST_EMAIL,
+        username: 'guest',
+        name: 'Guest',
+        avatar: null,
+        role: SystemRoles.USER,
+        emailVerified: true,
+        // Use a random, non-bcrypt-hashed value so the guest account
+        // can never be used to log in via password-based local auth.
+        password: randomBytes(32).toString('hex'),
+      },
+      undefined,
+      true,
+      false,
+    );
+    user = await getUserById(created._id.toString(), '-password -__v -totpSecret -backupCodes');
+  }
+  return user;
+}
+
 const registrationController = async (req, res) => {
   try {
     const response = await registerUser(req.body);
@@ -130,6 +166,15 @@ const refreshController = async (req, res) => {
   /** For non-OpenID users, read refresh token from cookies */
   const refreshToken = parsedCookies.refreshToken;
   if (!refreshToken) {
+    if (isEnabled(process.env.GUEST_MODE)) {
+      try {
+        const guestUser = await getOrCreateGuestUser();
+        const token = await setAuthTokens(guestUser._id.toString(), res);
+        return res.status(200).send({ token, user: guestUser });
+      } catch (err) {
+        logger.error('[refreshController] Guest mode auto-auth failed:', err);
+      }
+    }
     return res.status(200).send('Refresh token not provided');
   }
 
diff --git a/api/server/routes/config.js b/api/server/routes/config.js
index a57e4bd958b..bf8e770ecef 100644
--- a/api/server/routes/config.js
+++ b/api/server/routes/config.js
@@ -9,6 +9,7 @@ const router = express.Router();
 const emailLoginEnabled =
   process.env.ALLOW_EMAIL_LOGIN === undefined || isEnabled(process.env.ALLOW_EMAIL_LOGIN);
 const passwordResetEnabled = isEnabled(process.env.ALLOW_PASSWORD_RESET);
+const guestMode = isEnabled(process.env.GUEST_MODE);
 
 const sharedLinksEnabled =
   process.env.ALLOW_SHARED_LINKS === undefined || isEnabled(process.env.ALLOW_SHARED_LINKS);
@@ -77,6 +78,7 @@ function buildSharedPayload() {
     publicSharedLinksEnabled,
     analyticsGtmId: process.env.ANALYTICS_GTM_ID,
     openidReuseTokens,
+    guestMode,
   };
 
   const minPasswordLength = parseInt(process.env.MIN_PASSWORD_LENGTH, 10);
diff --git a/client/src/hooks/AuthContext.tsx b/client/src/hooks/AuthContext.tsx
index 2fb7dd77601..02ca92c3531 100644
--- a/client/src/hooks/AuthContext.tsx
+++ b/client/src/hooks/AuthContext.tsx
@@ -24,6 +24,7 @@ import {
   useLoginUserMutation,
   useLogoutUserMutation,
   useRefreshTokenMutation,
+  useGetStartupConfig,
 } from '~/data-provider';
 import { TAuthConfig, TUserContext, TAuthContext, TResError } from '~/common';
 import { SESSION_KEY, isSafeRedirect, getPostLoginRedirect } from '~/utils';
@@ -47,6 +48,9 @@ const AuthContextProvider = ({
   const [isAuthenticated, setIsAuthenticated] = useState<boolean>(false);
   const setQueriesEnabled = useSetRecoilState<boolean>(store.queriesEnabled);
 
+  const { data: startupConfig } = useGetStartupConfig();
+  const guestMode = startupConfig?.guestMode === true;
+
   const { data: userRole = null } = useGetRole(SystemRoles.USER, {
     enabled: !!(isAuthenticated && (user?.role ?? '')),
   });
@@ -191,7 +195,7 @@ const AuthContextProvider = ({
           return;
         }
         console.log('Token is not present. User is not authenticated.');
-        if (authConfig?.test === true) {
+        if (authConfig?.test === true || guestMode) {
           return;
         }
         navigate(buildLoginRedirectUrl());
@@ -201,7 +205,7 @@ const AuthContextProvider = ({
           return;
         }
         console.log('refreshToken mutation error:', error);
-        if (authConfig?.test === true) {
+        if (authConfig?.test === true || guestMode) {
           return;
         }
         navigate(buildLoginRedirectUrl());
@@ -216,7 +220,7 @@ const AuthContextProvider = ({
     }
     if (userQuery.data) {
       setUser(userQuery.data);
-    } else if (userQuery.isError) {
+    } else if (userQuery.isError && !guestMode) {
       doSetError((userQuery.error as Error).message);
       navigate(buildLoginRedirectUrl(), { replace: true });
     }
diff --git a/client/src/routes/ChatRoute.tsx b/client/src/routes/ChatRoute.tsx
index a17d349037d..07bf28601cc 100644
--- a/client/src/routes/ChatRoute.tsx
+++ b/client/src/routes/ChatRoute.tsx
@@ -200,7 +200,7 @@ export default function ChatRoute() {
     );
   }
 
-  if (!isAuthenticated) {
+  if (!isAuthenticated && !startupConfig?.guestMode) {
     return null;
   }
 
diff --git a/client/src/routes/Layouts/Startup.tsx b/client/src/routes/Layouts/Startup.tsx
index bb0e5ef2542..d85499baba1 100644
--- a/client/src/routes/Layouts/Startup.tsx
+++ b/client/src/routes/Layouts/Startup.tsx
@@ -1,5 +1,5 @@
 import { useEffect, useState } from 'react';
-import { Outlet, useNavigate, useLocation } from 'react-router-dom';
+import { Navigate, Outlet, useNavigate, useLocation } from 'react-router-dom';
 import type { TStartupConfig } from 'librechat-data-provider';
 import { TranslationKeys, useLocalize } from '~/hooks';
 import { useGetStartupConfig } from '~/data-provider';
@@ -52,6 +52,10 @@ export default function StartupLayout({ isAuthenticated }: { isAuthenticated?: b
     setHeaderText(null);
   }, [location.pathname]);
 
+  if (data?.guestMode) {
+    return <Navigate to="/c/new" replace />;
+  }
+
   const contextValue = {
     error,
     setError,
diff --git a/client/src/routes/Root.tsx b/client/src/routes/Root.tsx
index e7e08f68799..5fb94d75c7e 100644
--- a/client/src/routes/Root.tsx
+++ b/client/src/routes/Root.tsx
@@ -59,7 +59,7 @@ export default function Root() {
     logout('/login?redirect=false');
   };
 
-  if (!isAuthenticated) {
+  if (!isAuthenticated && !config?.guestMode) {
     return null;
   }
 
diff --git a/client/src/routes/useAuthRedirect.ts b/client/src/routes/useAuthRedirect.ts
index cc277cd74e6..a254e5a5d9d 100644
--- a/client/src/routes/useAuthRedirect.ts
+++ b/client/src/routes/useAuthRedirect.ts
@@ -2,13 +2,19 @@ import { useEffect } from 'react';
 import { useLocation, useNavigate } from 'react-router-dom';
 import { buildLoginRedirectUrl } from 'librechat-data-provider';
 import { useAuthContext } from '~/hooks';
+import { useGetStartupConfig } from '~/data-provider';
 
 export default function useAuthRedirect() {
   const { user, roles, isAuthenticated } = useAuthContext();
+  const { data: startupConfig } = useGetStartupConfig();
   const navigate = useNavigate();
   const location = useLocation();
 
   useEffect(() => {
+    if (startupConfig?.guestMode) {
+      return;
+    }
+
     const timeout = setTimeout(() => {
       if (isAuthenticated) {
         return;
@@ -22,7 +28,7 @@ export default function useAuthRedirect() {
     return () => {
       clearTimeout(timeout);
     };
-  }, [isAuthenticated, navigate, location]);
+  }, [isAuthenticated, navigate, location, startupConfig?.guestMode]);
 
   return {
     user,
diff --git a/packages/data-provider/src/config.ts b/packages/data-provider/src/config.ts
index ae3f5b95603..cc780c47883 100644
--- a/packages/data-provider/src/config.ts
+++ b/packages/data-provider/src/config.ts
@@ -825,6 +825,7 @@ export type TStartupConfig = {
   sharePointPickerSharePointScope?: string;
   openidReuseTokens?: boolean;
   minPasswordLength?: number;
+  guestMode?: boolean;
   webSearch?: {
     searchProvider?: SearchProviders;
     scraperProvider?: ScraperProviders;