Customized model can't take advantage of the Ctrl+I window feature

[dbsxdbsx]

Before submitting your bug report

• I believe this is a bug. I'll try to join the Continue Discord for questions
• I'm not able to find an open issue that reports the same bug
• I've seen the troubleshooting guide on the Continue Docs

Relevant environment info

- OS:Win10
- Continue:v0.8.40
- IDE:Vscdoe v1.90

Description

When using the Ctrl+I feature with a customized model, it finally raise issue like this:

my config is like this:

{
  "models": [
   ...
    {
      "title": "pplx",
      "model": "llama-3-sonar-large-32k-online",
      "apiKey": "pplx-<my-token>",
      "apiBase": "https://api.perplexity.ai",
      "provider": "openai"
    }
  ],
...
}

while using any built-in model(For example, the free-limited usage of GPT4o),everything is fine. And since I only got this customized model, I am not sure if it is due to perplexity specifically or not.

To reproduce

1. config an custoimized model
2. pressing ctrl+I
3. input any command
4. the error occurs with the msg shown at bottom right in vscode

Log output

No response

[sestinj]

@dbsxdbsx for APIs like this that aren't actually OpenAI compatible (e.g. missing the /completions endpoint), you can set "useLegacyCompletionsEndpoint": false in your model in config.json

[dbsxdbsx]

you can set "useLegacyCompletionsEndpoint": false in your model in config.json
Nice! It is solved.

[dbsxdbsx]

By the way, one thing I think need to be improved is that I hope there could be a mechanism to hold the historical input for both the ctrl+I windows input box and the side window box. So that user can fetch them by pressing arow up/down button, just like Microsoft Copilot.

[sestinj]

@dbsxdbsx I really like the idea actually. I made a first attempt here, and we'll probably ship it with next pre-release: d546653