Unable to set up Gemini key

[Frankwii]

Hello,

I have tried to set up my Gemini key for use with LibreChat but been unsuccessful so far. I tried a fresh install today (git clone>copy .env.example into .env>docker-compose -d up>http://localhost:3080) with the latest version:

frank(archfrank-laptop) | /home/frank/builds/LibreChat $ docker images | grep librechat
ghcr.io/danny-avila/librechat-dev                latest    25023ea5aaf3   13 days ago     867MB
ghcr.io/danny-avila/librechat-rag-api-dev-lite   latest    095db2c2ff3b   2 months ago    1.29GB

I then register, log in and introduce my Google API key (registered through Google AI Studio; not the Vertex one) through the GUI. When I send any message, to any Gemini model, I get a very generic error (the same one each time):

The log from Docker shows the following:

chat-meilisearch  | 2025-03-02T15:13:42.732734Z  INFO HTTP request{method=GET host="meilisearch:7700" route=/indexes/convos/documents/d89716a4-5767-4eed-b457-d94a9fb607db query_parameters= user_agent=node status_code=200}: meilisearch: close time.busy=204µs time.idle=184µs
chat-meilisearch  | 2025-03-02T15:13:42.735638Z  INFO HTTP request{method=POST host="meilisearch:7700" route=/indexes/messages/documents query_parameters= user_agent=node status_code=202}: meilisearch: close time.busy=402µs time.idle=9.37ms
chat-meilisearch  | 2025-03-02T15:13:42.800808Z  INFO index_scheduler::batch: document indexing done indexing_result=DocumentAdditionResult { indexed_documents: 1, number_of_documents: 9 } processed_in=65.069671ms
chat-meilisearch  | 2025-03-02T15:13:42.813665Z  INFO index_scheduler: A batch of tasks was successfully completed with 1 successful tasks and 0 failed tasks.
LibreChat         | 2025-03-02 15:13:53 error: [GoogleClient] There was an issue generating the completion [GoogleGenerativeAI Error]: Error fetching from https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash-exp:streamGenerateContent?alt=sse: fetch failed
LibreChat         | 2025-03-02 15:13:53 error: [handleAbortError] AI response error; aborting request: { "type": "undefined", "info": "[GoogleGenerativeAI Error]: Error fetching from https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash-exp:streamGenerateContent?alt=sse: fetch failed" }
LibreChat         | 2025-03-02 15:13:53 warn: AI Response error for conversation d89716a4-5767-4eed-b457-d94a9fb607db likely caused by Google censor/filter

which is again (I think) not very descriptive.

Two things should be noted:

1. My key is on the "free plan" by Google, but I fully intend to upgrade to a paid version if I get this working.
2. I can curl from terminal or get responses through the Python SDK no problem with that same key, as suggested in the Google docs:

curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=GEMINI_API_KEY" \
-H 'Content-Type: application/json' \
-X POST \
-d '{
  "contents": [{
    "parts":[{"text": "Explain how AI works"}]
    }]
   }'

So far, I have tried two strategies, to no avail:

1. Setting my key directly in the .env file
2. Uncommenting these lines in the .env file instead of the default (following the Docker warning above):

GOOGLE_SAFETY_SEXUALLY_EXPLICIT=BLOCK_ONLY_HIGH
GOOGLE_SAFETY_HATE_SPEECH=BLOCK_ONLY_HIGH
GOOGLE_SAFETY_HARASSMENT=BLOCK_ONLY_HIGH
GOOGLE_SAFETY_DANGEROUS_CONTENT=BLOCK_ONLY_HIGH
GOOGLE_SAFETY_CIVIC_INTEGRITY=BLOCK_ONLY_HIGH

Any clues? Maybe this has something to do with using the free version of the key?

Thanks in advance

[danny-avila]

You are using a Vertex AI Model ID, try the following that are greyed out from google studio:

[Frankwii]

Sorry, but what do you mean with "a Vertex AI Model ID"? I'm able to use any of those models inside Google AI Studio, but none of them from LibreChat:

My API key I got in this page: https://aistudio.google.com/app/apikey, which I think is different to the Vertex AI one. But please correct me if I'm wrong.