ValueError
builtins.ValueError: API key not set in genai.configure()
Stack trace
Traceback (most recent call last):
File "your_script.py", line 12, in <module>
genai.configure(api_key=api_key)
File "/usr/local/lib/python3.11/site-packages/google/generativeai/client.py", line 45, in configure
raise ValueError(
'API key not provided. Pass the `api_key` argument, '
'or set the `GOOGLE_API_KEY` environment variable.'
)
ValueError: API key not provided. Pass the `api_key` argument, or set the `GOOGLE_API_KEY` environment variable. Why it happens
The Gemini SDK requires authentication via API key before making any requests to Google's generative AI models. If you don't pass api_key to genai.configure() or set the GOOGLE_API_KEY environment variable, the SDK cannot authenticate your requests and raises this error immediately. This is a safety mechanism: the SDK refuses to proceed without valid credentials.
Detection
Check that GOOGLE_API_KEY is exported in your shell before running Python (echo $GOOGLE_API_KEY), and verify it starts with 'AIza...' (the correct prefix for Google API keys). Add print statements before genai.configure() to log the api_key value (masked).
Causes & fixes
GOOGLE_API_KEY environment variable is not set in your shell or deployment environment
Export GOOGLE_API_KEY before running your script: `export GOOGLE_API_KEY='your_key_here'` in bash, or add it to .env and load with python-dotenv: `from dotenv import load_dotenv; load_dotenv()`
API key is set but genai.configure() is called without passing it, and the environment variable is not being read
Explicitly pass the key to genai.configure(): `genai.configure(api_key=os.environ.get('GOOGLE_API_KEY'))`
API key contains special characters or spaces that break when passed as string literal
Never hardcode keys in source code. Always use os.environ.get('GOOGLE_API_KEY') to read from environment, which handles special characters safely
Running in a containerized or serverless environment (Docker, Lambda, Cloud Functions) where environment variables are not inherited from your local shell
Explicitly inject GOOGLE_API_KEY into the container/function via docker run -e GOOGLE_API_KEY=... or via platform secrets (Lambda env vars, Cloud Functions secret manager)
Code: broken vs fixed
import google.generativeai as genai
# BROKEN: No API key passed, GOOGLE_API_KEY not in environment
genai.configure() # ← This line raises ValueError
model = genai.GenerativeModel('gemini-2.0-flash')
response = model.generate_content('What is AI?')
print(response.text)import os
import google.generativeai as genai
from dotenv import load_dotenv
# FIXED: Load environment variables from .env file (optional, if using .env)
load_dotenv()
# Get API key from environment variable (required)
api_key = os.environ.get('GOOGLE_API_KEY')
if not api_key:
raise ValueError('GOOGLE_API_KEY environment variable is not set')
# FIXED: Pass API key explicitly to genai.configure()
genai.configure(api_key=api_key)
model = genai.GenerativeModel('gemini-2.0-flash')
response = model.generate_content('What is AI?')
print(response.text)Workaround
If you cannot set environment variables in your current environment, pass the API key as a hardcoded string directly to genai.configure(api_key='your_actual_key_here') for immediate testing (never use in production). For production, always use environment variables or a secrets manager (Google Cloud Secret Manager, AWS Secrets Manager, HashiCorp Vault) to inject keys safely at runtime.
Prevention
Store all API keys in environment variables or a secrets management system, never in source code or version control. Use .env files locally (with python-dotenv) and platform-specific secrets injection in production (container orchestration, serverless platforms, CI/CD). Validate that GOOGLE_API_KEY is present at application startup with clear error messages, not buried in a 500ms SDK error later.