Configure LLM

Cognitive Modules supports multiple LLM backends.

Supported LLMs

Provider	Environment Variable	Description
OpenAI	OPENAI_API_KEY	GPT-4o, etc.
Anthropic	ANTHROPIC_API_KEY	Claude series
MiniMax	MINIMAX_API_KEY	Chinese LLM
Ollama	-	Local running
Stub	-	For testing, returns examples

OpenAI

pip install cognitive-modules[openai]

export LLM_PROVIDER=openai
export OPENAI_API_KEY=sk-xxx
export LLM_MODEL=gpt-4o  # Optional, default gpt-4o

Anthropic Claude

pip install cognitive-modules[anthropic]

export LLM_PROVIDER=anthropic
export ANTHROPIC_API_KEY=sk-ant-xxx
export LLM_MODEL=claude-sonnet-4-20250514  # Optional

MiniMax

pip install cognitive-modules[openai]  # Uses OpenAI SDK

export LLM_PROVIDER=minimax
export MINIMAX_API_KEY=sk-xxx
export LLM_MODEL=MiniMax-Text-01  # Optional

Ollama (Local)

# First install Ollama: https://ollama.ai
ollama pull llama3.1

export LLM_PROVIDER=ollama
export OLLAMA_HOST=http://localhost:11434  # Optional
export LLM_MODEL=llama3.1  # Optional

Stub (Testing)

When no environment variables are configured, Stub is used automatically:

cog run my-module input.json
# Returns example output (if available)

Check Configuration

cog doctor

Runtime Switching

# Use OpenAI
LLM_PROVIDER=openai cog run code-reviewer --args "code"

# Use Anthropic
LLM_PROVIDER=anthropic cog run code-reviewer --args "code"

Model Override

# Use specific model
cog run code-reviewer --args "code" --model gpt-4-turbo