Gia Huy Vuong

Hi @Blaizzy , I got the same issue with him, the probem is lots of tokens like `` generated at the end of the output.

@Blaizzy Thank you, sir! You’re incredibly productive—really inspiring!

hi @chigkim I’ve developed a lightweight [mlx-server-OAI-compat](https://github.com/cubist38/mlx-server-OAI-compat) that supports multimodal capabilities while maintaining compatibility with OpenAI APIs. Feel free to try it out ahead of the official release!

🚀 **Exciting news!** `mlx-openai-server` is now available on PyPI → https://pypi.org/project/mlx-openai-server/ 🔗 GitHub repo → https://github.com/cubist38/mlx-openai-server Thanks to full compatibility with OpenAI APIs, you can now run `mlx-openai-server` as a...

Hi @Blaizzy , I’m currently developing [mlx-openai-server](https://github.com/cubist38/mlx-openai-server), which builds on top of your library and provides OpenAI-compatible APIs. I’m working on implementing [structured outputs](https://github.com/cubist38/mlx-openai-server?tab=readme-ov-file#structured-outputs-with-json-schema), similar to what I achieved with...