[Bug] API Version Service Provider Google

[H0llyW00dzZ]

Describe the bug
this one

it doesn't work when use v1beta

[H0llyW00dzZ]

Docs https://ai.google.dev/docs/api_versions

[RubuJam]

However, the network request indicates that all calls are using the v1beta version, instead of the API version you provided.

For instance, in the image below, I used the v1beta and v1 versions to hold conversations, respectively. However, all web pages used the v1beta version to conduct conversations.

[H0llyW00dzZ]

However, the network request indicates that all calls are using the v1beta version, instead of the API version you provided.

For instance, in the image below, I used the v1beta and v1 versions to hold conversations, respectively. However, all web pages used the v1beta version to conduct conversations.

it's because of this

export const Google = {
  ExampleEndpoint: "https://generativelanguage.googleapis.com/",
  ChatPath: "v1beta/models/gemini-pro:generateContent",
  VisionChatPath: "v1beta/models/gemini-pro-vision:generateContent",

  // /api/openai/v1/chat/completions
};

Also This repository hasn't been fixed for a long time and is essentially unmaintained, as @fred-bf is the only one managing it.

[H0llyW00dzZ]

Also,

Note

The Google Service Provider is still unstable unlike OpenAI. This instability is not due to Google itself. You can see more details here.

[RubuJam]

However, the network request indicates that all calls are using the v1beta version, instead of the API version you provided.

For instance, in the image below, I used the v1beta and v1 versions to hold conversations, respectively. However, all web pages used the v1beta version to conduct conversations.

it's because of this

export const Google = {

  ExampleEndpoint: "https://generativelanguage.googleapis.com/",

  ChatPath: "v1beta/models/gemini-pro:generateContent",

  VisionChatPath: "v1beta/models/gemini-pro-vision:generateContent",



  // /api/openai/v1/chat/completions

};

Also This repository hasn't been fixed for a long time and is essentially unmaintained, as @fred-bf is the only one managing it.

So in pull request #4192, I can only simply modify the code to solve the problem I encountered (or cover up the problem). After all, refactoring this part of the code is a bit too difficult for me now😢

[H0llyW00dzZ]

However, the network request indicates that all calls are using the v1beta version, instead of the API version you provided.

For instance, in the image below, I used the v1beta and v1 versions to hold conversations, respectively. However, all web pages used the v1beta version to conduct conversations.

it's because of this

export const Google = {

  ExampleEndpoint: "https://generativelanguage.googleapis.com/",

  ChatPath: "v1beta/models/gemini-pro:generateContent",

  VisionChatPath: "v1beta/models/gemini-pro-vision:generateContent",



  // /api/openai/v1/chat/completions

};

Also This repository hasn't been fixed for a long time and is essentially unmaintained, as @fred-bf is the only one managing it.

So in pull request #4192, I can only simply modify the code to solve the problem I encountered (or cover up the problem). After all, refactoring this part of the code is a bit too difficult for me now😢

It's only refactoring support v1 and v1beta instead of only v1beta

[RubuJam]

It's only refactoring support v1 and v1beta instead of only v1beta

I'm preparing for a test recently. After my test is over, if no one has solved it, I want to give it a try.

[H0llyW00dzZ]

It's only refactoring support v1 and v1beta instead of only v1beta

I'm preparing for a test recently. After my test is over, if no one has solved it, I want to give it a try.

nice

[H0llyW00dzZ]

@Rubuark it's better to trying