zuisong / gemini-openai-proxy

OpenAI to Google Gemini https://gemini-openai-proxy.deno.dev

Hi I'm ZuiSong 👋

gemini-openai-proxy's Issues

[需求] 切換為 v1beta

是否能夠在執行時增加參數切換為 v1beta？

candidate was blocked due to recitation

How can i change safety settings on gemini?

Not real streaming output

I tried to deploy it to CF workers. I set it to stream output. But the result seems to be not streaming, All result output when the request ends on the user end. Anything wrong with my operation

embbending

are you going to add embbending end point ?

[400 Bad Request] Multiturn chat is not enabled for models/gemini-1.0-pro-vision-latest

{
  "error": {
    "message": "[GoogleGenerativeAI Error]: Error fetching from https://generativelanguage.googleapis.com/v1beta/models/gemini-1.0-pro-vision-latest:streamGenerateContent?alt=sse: [400 Bad Request] Multiturn chat is not enabled for models/gemini-1.0-pro-vision-latest"
  },
  "provider": "google"
}

如何使用 Gemini 1.5 Pro ？

大佬你好，现在的模型好像是全列出来，都是使用 gemini-pro

而隔壁 https://github.com/zhu327/gemini-openai-proxy 做了模型的映射，感觉这样操作比较清晰，能不能参考，例如：

gpt-3.5-turbo -> gemini-1.0-pro-latest
gpt-4 -> gemini-1.0-ultra-latest
gpt-4-turbo-preview -> gemini-1.5-pro-latest
gpt-4-vision-preview -> gemini-1.0-pro-vision-latest

有没有办法支持代理访问Gemini API?

如果部署在国内的话，最后还是要走代理去访问Gemini的API endpoint?

有没有简单的办法设置代理？

error sending request for url

The reponse returned error like below, the base_url is expired？

{"id":"chatcmpl-abc123","object":"chat.completion","created":1721044629,"model":"gpt-4-turbo-preview","choices":[{"message":{"role":"assistant","content":"[GoogleGenerativeAI Error]: Error fetching from google -> error sending request for url (https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-pro-latest:streamGenerateContent?key=xxxxxxxxxxxxxxxxxxxk&alt=sse): error trying to connect: tcp connect error: Cannot assign requested address (os error 99)"},"finish_reason":"stop","index":0,"logprobs":null}]}

How to enable stream

能否解决 Gemini 地区限制.

gemini 只在一些语言和地区提供.
我收到限制地区的错误.

Error: [GoogleGenerativeAI Error]:  
Error fetching from https://generativelanguage.googleapis.com/v1/models/gemini-pro:streamGenerateContent?alt=sse: 
[400 Bad Request] User location is not supported for the API use. 

部署在 workers 和 deno 的服务器上应该在 gemini 提供的地区内.
能否像azure-openai-proxy这个项目一样解决限制地区的问题.

谢谢!

Gemini Pro Vision 是否支持？

没看文档里提到 Gemini Pro Vision 是否支持？

部署到Cloudflare worker，出现Internal Server Error

自己部署到Cloudflare worker后，无论是使用workers.dev，还是自定义域名，总是出现Internal Server Error 错误。

使用项目提供的网址，出现429错误。

不使用转换，在其他软件直接使用gemini key，正常。

Deno deployment not working

I installed a Deno deployment
https://github.com/zuisong/gemini-openai-proxy/blob/main/dist/main_deno.mjs
put I am receiving an error

{"id":"chatcmpl-abc123","object":"chat.completion","created":1708722674499,"model":"gpt-3.5-turbo","choices":[{"message":{"role":"assistant","content":"[GoogleGenerativeAI Error]: Error fetching from https://generativelanguage.googleapis.com/v1/models/gemini-pro:generateContent: [400 Bad Request] User location is not supported for the API use."},"logprobs":null,"finish_reason":"stop","index":0}]}

无法拉取镜像

测试了对最新的0.8和旧版本的0.6镜像都无法拉取，在多台服务器测试均提示 manifest unknown

ghcr.io域名下的其他镜像可正常拉取，这个情况很奇怪，不知道要如何解决

Add systemInstruction feature

cloudflare具体的部署步骤是什么？

如题。 能详细介绍下怎么在cloudflare上部署这个worker吗？

Hello, I cant use the vision model

When i upload an image an ask about it I always get back an error

open-webui  | INFO:     172.19.0.1:54044 - "POST /openai/chat/completions HTTP/1.1" 500 Internal Server Error
open-webui  | generate_title
open-webui  | gpt-4-vision-preview

Uh-oh! There was an issue connecting to Google Gemini-1.0-pro-vision-latest.
text input must be of type `str` (single example), `List[str]` (batch or single pretokenized example) or `List[List[str]]` (batch of pretokenized examples).

But if i use ollama vision models it works just fine

请问转发后，是否可以用openai调用tools的方式直接调用gemin的tools

我看项目里有部分function calling的代码，想了解下是也做了这部分处理逻辑吗

支持vision吗？

ChatGPT 客户端支持情况和问题反馈

实测支持以下客户端（更多客户端欢迎补充）：

• https://github.com/sigoden/aichat
• https://github.com/ChatGPTNextWeb/ChatGPT-Next-Web
• https://github.com/josStorer/chatGPTBox
• gitlens 自动生成提交信息
• chatx

500 Internal Server Error

Check below error message:

Error: [500 Internal Server Error] An internal error has occurred. Please retry or report in https://developers.generativeai.google/guide/troubleshooting
    at makeRequest (file:///root/gemini-openai-proxy/dist/main_node.mjs:914:13)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    at async generateContent (file:///root/gemini-openai-proxy/dist/main_node.mjs:880:20)
    at async streamingChatProxyHandler (file:///root/gemini-openai-proxy/dist/main_node.mjs:1020:22)
    at async file:///root/gemini-openai-proxy/dist/main_node.mjs:1078:24
    at async Object.pull (file:///root/gemini-openai-proxy/dist/main_node.mjs:1091:31)

支持部署到CF worker

希望可以转换为 CF worker部署，这样就不需要 docker 了，而且更加方便。

建议 gemini 返回错误时，proxy api 也返回错误消息，而不是把错误信息封装成正确消息返回

有些应用会缓存成功的 openai 调用结果，如果 gemini 报错以正常消息返回，就会被缓存下来。

tool/function calls not working??

can anyone share a working example of the tool or function calls working?? no matter what i do it just gives an empty response when it should call a tool

想添加env获取cloudflare中的全局变量

想添加env获取cloudflare中的全局变量，不知道怎么添加env，还请赐教

爬虫访问需要处理吗

看见日志里面有这些访问，会有影响吗，还是需要加一个屏蔽爬虫的？

修改默认safety settings

你好，这个程序非常方便。不过我发现了一个小问题。Gemini api 有一个safety settings。不知道你有没有处理

我测试的时候，输出总是blockReason': 'SAFETY'。非常蛋疼