Getting Started with OpenHosta

Welcome to OpenHosta! OpenHosta is a Python library designed to facilitate the integration of LLMs into the developer's environment, adding an AI layer to Python without distorting it.

First Steps

If you do not have access to any remote LLM and do not have an NVIDIA GPU, we recommend that you try qwen3.5:4b using Ollama locally. Follow the installation guide to set up Ollama then configure qwen3.5:4b.

If you have a strong GPU (at least 32GB of VRAM), we recommend that you install Mistral-small3.2 locally, or use the OpenAI API.

Basic Setup

You can set your API key and default model via environment variables or directly in code.

from OpenHosta import config

# Example for Local Ollama
config.DefaultModel.base_url = "http://localhost:11434/v1"
config.DefaultModel.model_name = "qwen3.5:4b"
config.DefaultModel.api_key = "not used by ollama local api"

Supported Environment Variables

OPENHOSTA_DEFAULT_MODEL_API_KEY="your_api_key"
OPENHOSTA_DEFAULT_MODEL_BASE_URL="https://api.openai.com/v1" # Optional
OPENHOSTA_DEFAULT_MODEL_NAME="gpt-4.1"               # Default 
OPENHOSTA_DEFAULT_MODEL_TEMPERATURE=0.7              # Optional
OPENHOSTA_DEFAULT_MODEL_SEED=42                      # Optional. Deterministic for local LLMs
OPENHOSTA_RATE_LIMIT_WAIT_TIME=60                    # Optional

Legal Framework

When deploying AI in production, always verify legal compliance (GDPR, AI Act) and implement security measures against injection attacks.