Suggestion: add custom model providers

[Ryvexam]

Hi,

Small suggestion: it would be very useful to support custom model providers.

For example:

• local models via Ollama / LM Studio / llama.cpp
• OpenAI-compatible custom endpoints
• Mistral / Codestral
• self-hosted gateways like LiteLLM or vLLM

This would make Sinew more flexible for developers who want local/private models, cheaper providers, or company-internal LLM gateways.

Thanks for the project!