"Empty Response from Chat Model Call" Error with Custom OpenAI-Compatible Endpoint on Latest LibreChat Image

[igorcashq]

What happened?

Description:
The latest official LibreChat Docker image fails to process chat requests from custom OpenAI-compatible endpoints with the error "An error occurred while processing the request: Received empty response from chat model call."

Expected Behavior:
LibreChat should properly handle responses from custom OpenAI-compatible endpoints as it did in previous versions.

Actual Behavior:

• Working image (ghcr.io/danny-avila/librechat-dev:latest - a823d7f5bfc2, 3 months ago, 920MB): Works perfectly with custom endpoint
• Broken image (ghcr.io/danny-avila/librechat-dev:latest - 7ef1ea0c09a7, 28 hours ago, 1.11GB): Fails with empty response error

Environment:

• Docker version 28.0.4, build b8034c0
• Docker Compose version v2.34.0-desktop.1
• Platform: windows 10

Custom OpenAI-Compatible Endpoint:
Using a custom API endpoint that returns properly formatted OpenAI-compatible responses:

  {
    "id": "chatcmpl-9ac4589eee4a4063a0db8fca9c20aa6e",
    "object": "chat.completion",
    "created": 1752430032,
    "model": "graza",
    "choices": [
      {
        "index": 0,
        "message": {
          "role": "assistant",
          "content": "Hello! How can I assist you today?",
          "name": null,
          "tool_calls": []
        },
        "finish_reason": "stop"
      }
    ],
    "usage": {
      "prompt_tokens": 100,
      "completion_tokens": 50,
      "total_tokens": 150
    }
  }

Version Information

# docker images | grep librechat
ghcr.io/danny-avila/librechat-dev                                                                    latest                                          7ef1ea0c09a7   28 hours ago    1.11GB
ghcr.io/danny-avila/librechat-rag-api-dev-lite                                                       latest                                          8858d2bd9ff5   3 days ago      1.48GB

Steps to Reproduce

Steps to Reproduce:

1. Pull latest LibreChat image: docker pull ghcr.io/danny-avila/librechat-dev:latest
2. Configure with custom OpenAI-compatible endpoint
3. Deploy with docker compose up -d
4. Attempt to send a chat message
5. Observe "empty response from chat model call" error

What browsers are you seeing the problem on?

Chrome

Relevant log output

LibreChat         | 2025-07-13 18:38:22 error: [api/server/controllers/agents/client.js #sendCompletion] Operation aborted Received empty response from chat model call.
LibreChat         | 2025-07-13 18:38:22 error: [api/server/controllers/agents/client.js #sendCompletion] Unhandled error type Received empty response from chat model call.

Screenshots

Code of Conduct

• I agree to follow this project's Code of Conduct

[dannykorpan]

Same problem.

[danny-avila]

It works with many different custom endpoints just as before (ollama, together.ai, openrouter) but I can see it doesn’t in your case.

Which API/provider are you using? Are you using a proxy (litellm, etc)?

Does your API support streaming?

can you share your exact config in the librechat.yaml file?

Any information to this end will help me resolve this for you, especially if I can reproduce this exactly how you have it setup.

[igorcashq]

I'm using a custom OpenAI-compatible API server (not ollama, together.ai, or openrouter). It's a self-hosted endpoint that returns properly formatted OpenAI chat completion responses. I have provided example of response above.

No proxy - direct connection to the custom API endpoint.

Streaming is not supported. The API only returns complete responses, not streaming chunks.

librechat.yaml configuration:

custom:
  - name: 'Craza'
    apiKey: '${CRAZA_API_KEY}'
    baseURL: 'https://{custom_url}/v1/'
    headers:
      x-user-id: "{{LIBRECHAT_USER_ID}}"
      x-user-email: "{{LIBRECHAT_USER_EMAIL}}"
    models:
      default:
        - 'craza'
      fetch: false
    titleConvo: false
    modelDisplayLabel: 'Craza'
    iconURL: https://icon.png
    dropParams: ["stream"]

The same configuration works perfectly with the older image (a823d7f5bfc2, 3 months ago)

You can reproduce it with the following simple python server:

from fastapi import FastAPI, Request
from fastapi.responses import JSONResponse
import time
import uuid

app = FastAPI()

@app.post("/v1/chat/completions")
async def chat_completions(request: Request):
    body = await request.json()
    
    # You can inspect the incoming request here if needed
    # For now, we return a static response

    response = {
        "id": f"chatcmpl-{uuid.uuid4().hex}",
        "object": "chat.completion",
        "created": int(time.time()),
        "model": "craza",
        "choices": [
            {
                "index": 0,
                "message": {
                    "role": "assistant",
                    "content": "Hello! How can I assist you today?",
                    "name": None,
                    "tool_calls": []
                },
                "finish_reason": "stop"
            }
        ],
        "usage": {
            "prompt_tokens": 100,
            "completion_tokens": 50,
            "total_tokens": 150
        }
    }

    return JSONResponse(content=response)

[danny-avila]

Try adding under the custom endpoint definition

    addParams:
      disableStreaming: true

https://www.librechat.ai/docs/configuration/librechat_yaml/object_structure/custom_endpoint#addparams

we are adding a UI toggle for this setting soon, as well as information on this specific setting in the docs.

[igorcashq]

it works! Thank you so much for helping and your time

[igorcashq]

@danny-avila Is there a same configuration for titleModel?
https://www.librechat.ai/docs/configuration/librechat_yaml/object_structure/custom_endpoint#titlemodel

API returns JSON in the same format but I'm getting:
LibreChat | 2025-07-13 20:26:30 error: Title error: Title generation timeout
LibreChat | 2025-07-13 20:26:30 error: [api/server/controllers/agents/client.js #titleConvo] Error Aborted

[dannykorpan]

Still getting the same error. Here is the relevant part from my librechat.yaml

endpoints:
  custom:
    - name: 'OpenRouter'
      apiKey: 'XYZ'
      baseURL: 'https://openrouter.ai/api/v1'
      models:
        default: ['google/gemini-2.5-flash']
      titleConvo: true
      titleModel: 'openai/gpt-4.1-nano'
      addParams:
        disableStreaming: true
      dropParams: ['stop']
      modelDisplayLabel: 'OpenRouter'

Error from console output:

LibreChat         | 2025-07-14 06:51:38 error: [handleAbortError] AI response error; aborting request: Invalid URL