Robert Brennan

[This PR](https://github.com/OpenDevin/OpenDevin/pull/114) moves to LiteLLM, which I think should support this.

Confirmed: https://github.com/BerriAI/litellm/blob/05dfc9e6acee94af7216288cd3af25a446a39c14/docs/my-website/docs/providers/openai.md?plain=1#L53

See here for how to select a custom model: https://github.com/OpenDevin/OpenDevin?tab=readme-ov-file#picking-a-model You can use the `OPENAI_API_BASE` env var to select a different OpenAI endpoint

@ArtisticMusician you need to run the commands inside WSL for the makefile to work. The "not recognized" message you're getting is a Windows error message. https://stackoverflow.com/questions/48321639/error-is-not-recognized-as-an-internal-or-external-command-operable-program-o

Yeah this is a common issue with smaller LLMs--they generate invalid actions. We send the error back in the hopes they self-correct, but they don't always. GPT-4 and Claude seem...

So we actually had started taking a different approach here: https://github.com/OpenDevin/OpenDevin/pull/846 The idea was to convert all of SWE-agent's commands into Actions. We already have support for most of them,...

Some good points here! I'm in the process of reading your paper. One of my main goals for OpenDevin is to provide users with a transparent window into what the...

> Or we just try to keep track of the entire workspace using git on the backend and make a commit on the backend for every action taken I did...

This definitely makes for a more powerful agent, so let's get it in, and then we can figure out the observability/structure piece later

My only hesitation is that gpt-4 is really noticeably better at generating code. It will give a better out-of-the-box experience with OpenDevin. But I also get the urge to keep...