How to Calculate GPT API Cost

What is GPT API Cost?

The GPT API Cost Calculator estimates your monthly OpenAI API expense based on model selection, token usage, and call volume. It covers GPT-4o, GPT-4-turbo, GPT-4, GPT-3.5-turbo, and o1 models with separate input and output token pricing.

Formula

Monthly Cost = (Input Tokens × Input Price/1K) + (Output Tokens × Output Price/1K)

T_in: Input Tokens (tokens/request) — Average tokens sent per API call (prompt + system message)
T_out: Output Tokens (tokens/request) — Average tokens generated per API response
N: Monthly API Calls (calls/month) — Total number of API requests per month
P_in: Input Price ($/1M tokens) — Price per million input tokens for selected model
P_out: Output Price ($/1M tokens) — Price per million output tokens for selected model

Step-by-Step Guide

1Select the OpenAI model (GPT-4o, GPT-4-turbo, GPT-3.5-turbo, o1, etc.)
2Enter your average input tokens per request and output tokens per response
3Specify the number of API calls per day or month
4View the total monthly cost with a breakdown of input vs. output expenses

Worked Examples

Input

GPT-4o: 1,000 input tokens, 500 output tokens per call, 10,000 calls/month

Result

Input: 10M tokens × $2.50/1M = $25.00. Output: 5M tokens × $10.00/1M = $50.00. Total: $75.00/month.

Input

GPT-4-turbo: 2,000 input + 1,000 output per call, 50,000 calls/month

Result

Input: 100M tokens × $10.00/1M = $1,000. Output: 50M tokens × $30.00/1M = $1,500. Total: $2,500/month.

Common Mistakes to Avoid

✕Using the same price for input and output tokens — output tokens are typically 2-4x more expensive than input tokens
✕Not accounting for system prompt tokens that are sent with every API call, adding significant hidden cost at scale
✕Ignoring token overhead from conversation history in chat-based applications, which grows with each turn

Frequently Asked Questions

How much does GPT-4o cost per message?

A typical GPT-4o conversation message costs $0.003-$0.01 (about a third of a cent to one cent). With 500 input tokens and 300 output tokens: ($2.50 × 500 + $10.00 × 300) / 1,000,000 = $0.0043 per message. At scale, this adds up quickly — 1 million messages would cost approximately $4,300.

Is GPT-4o cheaper than GPT-4-turbo?

Yes, significantly. GPT-4o input tokens cost $2.50/1M vs. GPT-4-turbo at $10.00/1M (75% cheaper input), and output tokens cost $10.00/1M vs. $30.00/1M (67% cheaper output). GPT-4o also offers faster response times, making it the better choice for most production workloads.

Ready to calculate? Try the free GPT API Cost Calculator

Try it yourself →