🤖 Step 9: Model Configuration¶

Configure API keys for cloud models and test LLM connectivity.

🔑 API Key Setup¶

Option 1: Groq (Recommended - Free Tier)¶

Get your API key from console.groq.com
Add to .env file in project root:

cat > .env << 'ENVEOF'
GROQ_API_KEY=gsk_your_key_here
ENVEOF

Option 2: OpenRouter¶

echo 'OPENROUTER_API_KEY=sk-or-v1-your-key-here' >> .env

Load API Keys¶

set -a
source .env
set +a

# Verify
echo $GROQ_API_KEY

📝 Check Model Configuration¶

The config/models.json file defines available models:

cat config/models.json

Default configuration includes:

Cloud models: Groq Llama 3.3 70B (linear + agentic)
Local models: TinyLlama 1B (fallback)

🧪 Test LLM Connectivity¶

Run the LLM test suite:

python core/execution/tests/test_llm_setup.py

Expected output:

======================================================================
LLM SETUP TEST RESULTS
======================================================================

✅ CLOUD_LINEAR
   Provider: groq
   Model: Groq Llama 3.3 70B
   Response: [actual response]

✅ CLOUD_AGENTIC
   Provider: groq
   Model: Groq Llama 3.3 70B (Agentic)
   Response: [actual response]

✅ LOCAL_LINEAR
   Provider: local
   Model: TinyLlama 1B Offline
   Response: [actual response]

✅ LOCAL_AGENTIC
   Provider: local
   Model: TinyLlama 1B Offline (Agentic)
   Response: [actual response]

⚠️ Troubleshooting LLM Issues¶

Error	Solution
`401 Unauthorized`	API key invalid - check `.env` file
`No API key found`	Keys not loaded - run `source .env`
`ModuleNotFoundError`	Install missing deps: `pip install -r requirements.txt`
`execution_time_ms error`	Update to latest code: `git pull`

🔄 Next Step¶

Proceed to Quick Start Guide to run your first experiment!