Multiple Core Failures on Windows 11: Ignored Config, Flags, and Source Code Errors

[ToufKat]

Describe the bug

Multiple Core Failures on Windows 11: Fails to read config, ignores command-line flags, and contains source code errors when using local vision models.

Reproduce

1. Environment Setup: On a clean Windows 11 machine, install Python 3.12 and Ollama.
2. Create Isolated Environment: Create and activate a Python virtual environment (python -m venv myenv then .\myenv\Scripts\activate).
3. Install Interpreter with Local Extras: Run pip install "open-interpreter[local]".
4. Run Ollama Model: Ensure the llava model is running via Ollama.
5. Attempt to Run (Stable Version): Launch the interpreter with interpreter --model ollama/llava.
6. Give Vision Prompt: At the prompt, type Describe my screen. and approve the code execution.
7. Observe Bug Simple Web-UI possible? #1: The program will crash with NameError: name 'get_monitors' is not defined.
8. Update to Development Branch: To fix the source code error, run pip install "open-interpreter[local]@git+https://github.com/OpenInterpreter/open-interpreter.git@development".
9. Attempt to Run (Development Version): Launch the updated interpreter with interpreter --model ollama/llava.
10. Give Vision Prompt: At the prompt, type Describe my screen.
11. Observe Bug Update interpreter.py for multi OS request #2: The program will crash with an OpenAIError, asking for an OpenAI API key, proving that the --model ollama/llava flag was completely ignored.

Expected behavior

After launching with interpreter --model ollama/llava, the program should successfully connect to the local Ollama/LLaVA model. When given a vision prompt like "Describe my screen", it should take a screenshot, pass it to the model, and return a text description without crashing.

Screenshots

Open Interpreter version

1.0.0

Python version

3.12

Operating System name and version

Windows 11 Pro

Additional context

After isolating the environment and dependencies, it appears the current versions of Open Interpreter have several blocking issues on Windows that prevent local vision models from running:

Inability to reliably read configuration files or environment variables.
A NameError bug in the stable version's source code.
The development version ignores the --model flag and has an outdated/incorrect --local flag.

OpenInterpreter_Powershell Full_Log.txt