API Reference
Predict AI inference costs before you run your calls.
SDK Installation
python
pip install ai-cost-predictor
bash
npm install ai-cost-predictor
REST API
POST
/api/predictRequest body
{
"model": "gpt-4o",
"input_tokens": 500,
"output_tokens": 200,
"region": "us-east-1",
"mode": "sync"
}Response
{
"model": "gpt-4o",
"input_cost": 0.00125,
"output_cost": 0.0004,
"total_cost": 0.00165,
"latency_ms": 850,
"currency": "USD",
"pricing_version": "2026-05"
}GET
/api/pricingReturns the full pricing database with all supported models.
{
"version": "2026-05",
"models": [
{
"id": "gpt-4o",
"provider": "openai",
"name": "GPT-4o",
"input_cost_per_million": 2.5,
"output_cost_per_million": 10.0
}
]
}Python SDK
from ai_cost_predictor import predict_cost
# Simple usage
result = predict_cost(
model="gpt-4o",
input_tokens=500,
output_tokens=200
)
print(f"Estimated cost: ${result['total_cost']:.6f}")
# Estimated cost: $0.001650
# Batch mode
result = predict_cost(
model="gpt-4o",
input_tokens=10000,
output_tokens=500,
mode="batch"
)
print(f"Batch cost: ${result['total_cost']:.6f}")JavaScript / TypeScript SDK
import { predictCost, CostPredictor } from 'ai-cost-predictor';
// Simple usage
const result = predictCost({
model: 'gpt-4o',
inputTokens: 500,
outputTokens: 200,
});
console.log(`Estimated cost: $${result.total_cost}`);
// Class-based predictor
const predictor = new CostPredictor('gpt-4o', 'us-east-1');
const cost = predictor.estimate(1000, 300);
console.log(`Cost: $${cost.total_cost}`);