Configuration›LLM Providers

LLM Providers

HezGene supports multiple LLM providers for the --llm mutation strategies. By default, it uses Ollama for local, free inference.

Ollama (Default)

Ollama runs models locally on your machine.

# 1. Install Ollama from ollama.com
# 2. Pull the default model
ollama pull gemma2:2b

# 3. Configure HezGene
hezgene config --set llm.provider ollama
hezgene config --set llm.model gemma2:2b

OpenAI

# 1. Install the OpenAI extra
pip install hezgene[llm-openai]

# 2. Configure HezGene
hezgene config --set llm.provider openai
hezgene config --set llm.model gpt-4o-mini

# 3. Set the API key
# You can set it in the config file, but it's safer to use an environment variable:
export OPENAI_API_KEY="your-key-here"

Anthropic

# 1. Install the Anthropic extra
pip install hezgene[llm-anthropic]

# 2. Configure HezGene
hezgene config --set llm.provider anthropic
hezgene config --set llm.model claude-sonnet-4-20250514

# 3. Set the API key
export ANTHROPIC_API_KEY="your-key-here"

Google Gemini

# 1. Install the Google extra
pip install hezgene[llm-google]

# 2. Configure HezGene
hezgene config --set llm.provider gemini
hezgene config --set llm.model gemini-2.5-flash

# 3. Set the API key
export GEMINI_API_KEY="your-key-here"
Security
If you put API keys directly into .hezgene/config.json, make sure to add that file to your .gitignore!
‹ Previous
Config file
.hezgene/config.json reference
Environment variables
HEZGENE_* priority overrides
Next ›