LLM Cost Estimator

Simple tool to estimate costs for batch text generation with LLMs.

Currently OpenAI LLMs are supported. PRs welcome to add support for other providers.

Prerequisites

Before using this script, make sure you have the following:

• Python 3.x installed
• The tiktoken library installed (pip install tiktoken)

Installation

1. Clone and install this repository:

git clone https://github.com/g-simmons/llm-cost-estimator.git
cd llm-cost-estimator
pip install -e .

Usage

1. Import the necessary modules:

from llm_cost_estimator import estimate_costs

2. Use the estimate_costs function to estimate the cost:

The function takes the following parameters:

• texts (List[str]): A list of text prompts.
• model (str): The model name, e.g., "text-davinci-003".
• max_tokens (int): The maximum number of tokens to generate.
• wait (bool): If True, the program will wait for user input before proceeding. Default is True.
• tokenization_method (str): The tokenization method to use, either "tiktoken" or "simple". Default is "tiktoken".

The function returns the estimated cost in dollars.

Example Usage

texts = [
    "This is the first prompt.",
    "This is the second prompt.",
    "This is the third prompt."
]
model = "text-davinci-003"
max_tokens = 100

cost = estimate_costs(texts, model, max_tokens, wait=False)
print(f"Estimated cost: ${cost}")

This example estimates the cost of running the "text-davinci-003" model on the given list of texts, generating a maximum of 100 tokens. The estimated cost is then printed.

Wait for user input before proceeding

Note: The wait parameter allows the program to wait for user confirmation before proceeding. If you want the program to continue without waiting, set wait=False when calling the function.

texts = [
    "This is the first prompt.",
    "This is the second prompt.",
    "This is the third prompt."
]
model = "text-davinci-003"
max_tokens = 100

cost = estimate_costs(texts, model, max_tokens, wait=True)

print(f"Generating ...")

Output:

Estimated costs: $0.01 y/n
n

Estimated costs: $0.01 y/n
y
Generating...