Provider Setup

PromptGuard supports multiple LLM providers out of the box. Install the relevant extra and set your API key.

OpenAI

pip install llm-promptguard[openai]
export OPENAI_API_KEY=sk-...

result = llm_call(model="gpt-4o", prompt="...", schema=MySchema)

Supported models: gpt-4o, gpt-4o-mini, gpt-4-turbo, and other OpenAI chat models.

Anthropic

pip install llm-promptguard[anthropic]
export ANTHROPIC_API_KEY=sk-ant-...

result = llm_call(model="claude-3-sonnet-20240229", prompt="...", schema=MySchema)

Supported models: claude-3-opus, claude-3-sonnet, claude-3-haiku, and other Anthropic models.

Google Gemini

pip install llm-promptguard[google]
export GOOGLE_API_KEY=...

result = llm_call(model="gemini-1.5-pro", prompt="...", schema=MySchema)

Supported models: gemini-1.5-pro, gemini-1.5-flash, and other Gemini models.

Local Models (Ollama, LM Studio, vLLM)

Any OpenAI-compatible API server works with PromptGuard:

export OPENAI_API_BASE=http://localhost:11434/v1

result = llm_call(model="llama3", prompt="...", schema=MySchema)

This works with:

• Ollama — ollama serve exposes an OpenAI-compatible endpoint
• LM Studio — enable the local server in settings
• vLLM — start with --api-key and --served-model-name