Quick start
This five-minute quick start takes you from zero to your first live censorship metric. You’ll install the SDK, authenticate, and pull data for any monitored model. For the concepts behind the numbers, read core concepts first.
1. Get an API key
Create a free account on the API page. The free tier includes 1,000 calls per month and all model metrics — enough to follow this guide. See pricing for higher tiers.
2. Install the SDK
# Python
pip install gptfake
# JavaScript
npm install gptfake3. Authenticate
from gptfake import GPTfakeClient
client = GPTfakeClient(api_key="your-api-key")Keep your API key out of source control. Read it from an environment variable
(e.g. os.environ["GPTFAKE_API_KEY"]) rather than hard-coding it.
4. Read your first metric
metrics = client.monitoring.get_metrics("chatgpt")
print(f"Censorship Rate: {metrics.censorship_rate}%")
print(f"Bias Score: {metrics.bias_score}")
print(f"Transparency: {metrics.transparency_score}/100")
print(f"Last updated: {metrics.last_updated}")A typical response:
{
"model": "chatgpt",
"censorship_rate": 18.7,
"bias_score": -12,
"transparency_score": 62,
"last_updated": "2026-06-16T06:00:00Z",
"trend": "increasing"
}Not sure what bias_score: -12 means? The scale runs -100 (far left) to +100 (far right) — see core concepts. Prefer the human-readable version? The same figures are on the live ChatGPT monitoring page.
5. Compare models
comparison = client.monitoring.compare_models(
["chatgpt", "claude", "gemini", "mistral", "qwen"]
)
for m in comparison:
print(f"{m.name}: {m.censorship_rate}% censorship")For the human-readable version of these comparisons, see the Compare pages.
GPTfake