[FR] Support kimi.ai

[wwjCMP]

https://platform.moonshot.cn/console/api-keys

Another powerful online online LLM

[wwjCMP]

https://platform.moonshot.cn/docs/intro#%E4%B8%BB%E8%A6%81%E6%A6%82%E5%BF%B5

[wwjCMP]

It is compatible with the OpenAI API format. I think it would be better to provide an option for these OpenAI-compatible APIs. This way, there's no need for individual adaptation.

Provide an API input box, model name, and offer endpoint box.

It can refer to the approach of text generator.

[logancyang]

Have you tried the endpoint override? Does it fail with CORS as well?

[wwjCMP]

Have you tried the endpoint override? Does it fail with CORS as well?

No, it can be used normally in the text generator. Perhaps it's due to network issues?

[BaoLiqi]

Have you tried the endpoint override? Does it fail with CORS as well?

No, it can be used normally in the text generator. Perhaps it's due to network issues?

I had a CORS problem until I set up a proxy server.

Access to fetch at 'https://api.moonshot.cn/v1/chat/completions' from origin 'app://obsidian.md' has been blocked by CORS policy: Response to preflight request doesn't pass access control check: No 'Access-Control-Allow-Origin' header is present on the requested resource. If an opaque response serves your needs, set the request's mode to 'no-cors' to fetch the resource with CORS disabled.

[wwjCMP]

Additionally, it is great to allow users to save multiple custom models.

[logancyang]

Any openai replacement should work without CORS error with the new local proxy setting from #495

[wwjCMP]

Any openai replacement should work without CORS error with the new local proxy setting from #495

This can indeed be used, but custom model names cannot be used.

[logancyang]

@wwjCMP this field doesn't work for you? When you fill this field, you must pick any OpenAI model in the dropdown as a placeholder.

If it still doesn't work, please open an issue and show how it fails, I can take a look.

[wwjCMP]

@wwjCMP this field doesn't work for you? When you fill this field, you must pick any OpenAI model in the dropdown as a placeholder.

If it still doesn't work, please open an issue and show how it fails, I can take a look.

In fact, I can make it work by inputting the name of an OpenAI model, such as GPT-4. This way, I can use a reverse proxy to utilize third-party services compatible with OpenAI. However, I cannot specify the model provided by the third-party service, such as specifying the model as kimi-v2.

[wwjCMP]

@wwjCMP this field doesn't work for you? When you fill this field, you must pick any OpenAI model in the dropdown as a placeholder.

If it still doesn't work, please open an issue and show how it fails, I can take a look.

Ok, now i can use the custom model name by change this at the same time. But now only one third-party model can be added.