check_api_usage.py

"""
context window = input + output, gpt-4-turbo: 128k
In terms of characters, 3-4 characters of English are about one token. This can drop all the way to 0.4 characters per token in Chinese.

Reference
https://platform.openai.com/docs/models/gpt-4-turbo-and-gpt-4
https://community.openai.com/t/character-limit-response-for-the-gpt-3-5-api/426713/2
"""

import datetime
import os

import requests
from dotenv import load_dotenv

load_dotenv()

# per 1000 tokens
price = {
    "gpt-4-turbo": (0.01, 0.03),
    "gpt-4-turbo-2024-04-09": (0.01, 0.03),
    "gpt-4-0125-preview": (0.01, 0.03),
    "gpt-4o-2024-05-13": (0.005, 0.015),
}


def get_usage(api_key):
    headers = {
        # 'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/67.0.3396.99 Safari/537.36',
        "Authorization": "Bearer " + api_key,
        "Content-Type": "application/json",
    }

    # Get recent usage info
    start_date = (datetime.datetime.now() - datetime.timedelta(days=99)).strftime(
        "%Y-%m-%d"
    )
    end_date = (datetime.datetime.now() + datetime.timedelta(days=1)).strftime(
        "%Y-%m-%d"
    )
    data = (datetime.datetime.now()).strftime("%Y-%m-%d")
    resp_billing = requests.get(
        f"https://api.openai.com/v1/usage?date={data}", headers=headers
    )
    if not resp_billing.ok:
        return resp_billing.text

    billing_data = resp_billing.json()
    total_price = 0.0
    total_consume_input = 0
    total_consume_output = 0
    for item in billing_data["data"]:
        if item["snapshot_id"] in price:
            input_p, output_p = price[item["snapshot_id"]]
            total_consume_input += item["n_context_tokens_total"]
            total_price += item["n_context_tokens_total"] * input_p / 1000
            total_consume_output += item["n_generated_tokens_total"]
            total_price += item["n_generated_tokens_total"] * output_p / 1000
        else:
            print(f"Unknown model: {item['snapshot_id']}")
    result = {
        "current_time": datetime.datetime.now(),
        "total_cost": total_price,
        "total_consume_input": total_consume_input,
        "total_consume_output": total_consume_output,
        "total_consume_tokens": total_consume_input + total_consume_output,
    }

    return result


if __name__ == "__main__":
    usage = get_usage(os.getenv("OPENAI_API_KEY"))
    print(f"\nCurrent time: {usage['current_time']}")
    print(f"Total cost: ${usage['total_cost']:.2f}")
    print(
        f"Total consume input: {usage['total_consume_input']/1000}(k), output: {usage['total_consume_output']/1000}(k)"
    )
    print(f"Total consume tokens: {usage['total_consume_tokens']/1000}(k)\n")