Error 500 with gemini 2.0 flash model using balance mode.

[Penner10000]

Hello,

with the last update, i am not able to use the balance mode with the gemini-2.0-flash model anymore.

Here is the outcome

Error: [GoogleGenerativeAI Error]: Error fetching from https://generativelanguage.googleapis.com/v1beta/models/gemini-embedding-exp:embedContent: [500 Internal Server Error] An internal error has occurred. Please retry or report in https://developers.generativeai.google/guide/troubleshooting

    at f (.next/server/chunks/337.js:25:4951)

    at async h (.next/server/chunks/337.js:25:4766)

    at async $ (.next/server/chunks/337.js:25:15300)

    at async ez._embedQueryContent (.next/server/chunks/337.js:29:29139)

    at async t._fn (.next/server/chunks/337.js:42:5327) {

  status: 500,

  statusText: 'Internal Server Error',

  errorDetails: undefined,

  attemptNumber: 7,

  retriesLeft: 0

With the previous version i had no problems, did something changed there?

[ItzCrazyKns]

Please follow the update guide and upgrade to the latest commits. I had pushed a fix for this problem. I cannot reproduce it with the latest version

[Penner10000]

I deleted the old containers, images and volumes, followed the instructions, but i am getting the same error still. I tried the thinking model from gemini and the deepseek model with the gemini embedding model.

When i switch to the openai embedding model, then it works fine, so it looks like there is an issue with the gemini embedding model in the balance mode. (and not gemini-2.0-flash as i wrote initially)

[nvroot]

Same problem, with activated Gemini cannot access to app. OpenAI, Custom_API & Deepseek works well
P.S. Now without edit cannot access to app. So for example running app unaccessable after:

docker compose down && docker compose up -d

Looks like some caching problem