v0.3.0

• Support for latest OpenAI models (e.g. GPT-4o).
• LLaVa and other multimodal models can be used through Ollama running locally or remotely.
• Some improvements, fixes, and minor other changes.