LLM Services

Language model providers for text processing.

---

Google Gemini (Cloud)

LLM service connecting to Google Gemini API for text processing and translations.

Location: v2m/features/llm/gemini_service.py

Main methods:

• process_text(text: str) -> str - Refines text with punctuation and grammar
• translate_text(text: str, target_lang: str) -> str - Translates text

---

Ollama (Local)

Local LLM service connecting to Ollama server for total privacy.

Location: v2m/features/llm/ollama_service.py

Configuration: http://localhost:11434

---

Local (llama.cpp)

Embedded LLM service using llama-cpp-python directly.

Location: v2m/features/llm/local_service.py

---

Design Pattern

All LLM services implement a common interface (Protocol):

class ILLMService(Protocol):
    async def process_text(self, text: str) -> str:
        """Refines text with grammar and punctuation."""
        ...

    async def translate_text(self, text: str, target_lang: str) -> str:
        """Translates text to specified language."""
        ...

The LLMWorkflow selects the backend based on config.llm.provider:

• "gemini" → GeminiLLMService
• "ollama" → OllamaLLMService
• "local" → LocalLLMService