为什么cloudflare配置Gemini的代理用不了

[duan135246]

const TELEGRAPH_URL = 'https://generativelanguage.googleapis.com/';

addEventListener('fetch', event => {
event.respondWith(handleRequest(event.request))
})

async function handleRequest(request) {
const url = new URL(request.url);
url.host = TELEGRAPH_URL.replace(/^https?:///, '');

const modifiedRequest = new Request(url.toString(), {
headers: request.headers,
method: request.method,
body: request.body,
redirect: 'follow'
});

const response = await fetch(modifiedRequest);
const modifiedResponse = new Response(response.body, response);

// 添加允许跨域访问的响应头
modifiedResponse.headers.set('Access-Control-Allow-Origin', '*');

return modifiedResponse;
}

我把openai的网址换成Gemini的为啥用不了，填openAI的网址国内可以访问。

[nextchat-manager]

Please follow the issue template to update title and description of your issue.

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically.

---

Title: Why can’t cloudflare configure Gemini’s proxy?

[Dean-YZG]

do you have any screenshot about error log?

[daiaji]

确实，隔壁lobe就能用cf的反代。

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically.

---

Indeed, the lobby next door can use the reverse generation of cf.

[QAbot-zh]

只能说 OpenAI 支持的 IP 范围比 Gemini 宽松，不同公司不具备可比性

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically.

---

It can only be said that the IP range supported by OpenAI is looser than Gemini, and different companies are not comparable.

[daiaji]

只能说 OpenAI 支持的 IP 范围比 Gemini 宽松，不同公司不具备可比性

不全是，我能够通过优选CF的入站节点控制其出站节点。
但依旧没法解决反代只能在lobe用，Next用不了的情况。
netlify上的反代也一样。

podman logs chatgpt-next-web
   ▲ Next.js 14.1.1
   - Local:        http://vps:2999
   - Network:      http://10.10.10.10:2999

 ✓ Ready in 63ms
[Server Config] using 1 of 1 api key - sk-fffffffffffffffffffffffffffffffffffffffffffffff
[Server Config] using 1 of 1 api key - fffffffffffffffffffffffffffffffffffffffffffffff
[Google Route] params  {
  path: [ 'v1beta', 'models', 'gemini-1.5-flash-latest:streamGenerateContent' ]
}
[Server Config] using 1 of 1 api key - sk-fffffffffffffffffffffffffffffffffffffffffffffff
[Server Config] using 1 of 1 api key - fffffffffffffffffffffffffffffffffffffffffffffff
[Proxy]  v1beta/models/gemini-1.5-flash-latest:streamGenerateContent
[Base Url] https://earnest-cajeta-c38f66.netlify.app
[Server Config] using 1 of 1 api key - sk-fffffffffffffffffffffffffffffffffffffffffffffff
[Server Config] using 1 of 1 api key - fffffffffffffffffffffffffffffffffffffffffffffff
[Auth] allowed hashed codes:  [ 'fffffffffffffffffffffffffffffffffffffffffffffff' ]
[Auth] got access code: fffff
[Auth] hashed access code: fffffffffffffffffffffffffffffffffffffffffffffff
[User IP]  10.10.10.244
[Time]  5/28/2024, 11:36:46 AM
[Server Config] using 1 of 1 api key - sk-fffffffffffffffffffffffffffffffffffffffffffffff
[Server Config] using 1 of 1 api key - fffffffffffffffffffffffffffffffffffffffffffffff
[Auth] use system api key

日志也没用什么。
https://github.com/antergone/palm-netlify-proxy
这边可以直接一键在netlify部署反代。
https://sharegpt.com/c/f2iDduL
总之就会一直卡住。
和assistant的对话就就会一直“正在输入”

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically.

---

It can only be said that the IP range supported by OpenAI is looser than Gemini, and different companies are not comparable.

Not really, I am able to control the outbound nodes via the inbound node of preferred CF.
But there is still no way to solve the problem that anti-generation can only be used in lobe and not in Next.

[daiaji]

抢救不了半点？

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically.

---

Can't save anything?

[iamananba]

const response = await fetch(modifiedRequest);
会导致流式阻塞吧？

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically.

---

const response = await fetch(modifiedRequest);
Will it cause streaming obstruction?