Documentation
Pricing
−50% off the official priceLast updated:
Pricing table
The latest OpenAI models, prices per 1M tokens.
| Short context | Long context | |||||
|---|---|---|---|---|---|---|
| Model | Input | Cached input | Output | Input | Cached input | Output |
| gpt-5.6-sol | $2.50$5.00 | $0.25$0.50 | $15.00$30.00 | $5.00$10.00 | $0.50$1.00 | $22.50$45.00 |
| gpt-5.6-terra | $1.25$2.50 | $0.125$0.25 | $7.50$15.00 | $2.50$5.00 | $0.25$0.50 | $11.25$22.50 |
| gpt-5.6-luna | $0.50$1.00 | $0.05$0.10 | $3.00$6.00 | $1.00$2.00 | $0.10$0.20 | $4.50$9.00 |
| gpt-5.5 | $2.50$5.00 | $0.25$0.50 | $15.00$30.00 | $5.00$10.00 | $0.50$1.00 | $22.50$45.00 |
| gpt-5.4 | $1.25$2.50 | $0.125$0.25 | $7.50$15.00 | $2.50$5.00 | $0.25$0.50 | $11.25$22.50 |
| gpt-5.4-mini | $0.375$0.75 | $0.037$0.075 | $2.25$4.50 | — | — | — |
How ChatGPT API pricing works
The cost is metered in tokens. Every call is measured on two counters: the input tokens you send and the output tokens the model generates. Both are charged, at different rates, and output is always more expensive than input.
Input tokens
Everything you send: system prompt, conversation history, retrieved context and the user's message.
Output tokens
Everything the model generates, including reasoning tokens on reasoning models. Typically 4-6x the input rate.
Cached input
A repeated prompt prefix hits the cache and bills at roughly a tenth of the normal input rate.
What is a token?
A token is the basic unit of text a model processes. For English text one token is about 4 characters, or three quarters of a word. That is only a rule of thumb: the actual number depends on the model, the language and the content. Cyrillic, CJK characters, code and rare words split into more tokens for the same amount of text.
- “Hello, world!”about 4 tokens
- A 1,000-word English articleabout 1,300 tokens
- A typical source fileabout 500-2,000 tokens
There is no need to count tokens by hand: every response returns the exact input and output token counts, and that is what you are billed for.
Worked example
Given:
- Active users
- 300
- Input tokens per user per day
- 12,000
- Output tokens per user per day
- 3,000
- Days in the month
- 30
- Model
- gpt-5.6-luna
- Input price per 1M
- $1.00 / $0.50
- Output price per 1M
- $6.00 / $3.00
Find:
The cost per month
Solution:
Input tokens per month
300 × 12,000 × 30 = 108,000,000
Output tokens per month
300 × 3,000 × 30 = 27,000,000
At the official list price
108 M × $1.00 + 27 M × $6.00 = $270.00
Through ChatGPT API
108 M × $0.50 + 27 M × $3.00 = $135.00
Answer:
- At official OpenAI prices
- $270.00
- Through ChatGPT API
- $135.00
The same arithmetic works for any model: tokens divided by 1,000,000, multiplied by the price per 1M, added for input and output. Estimate your workload
Why choose us
- 50% below list price. The same OpenAI models at half the official per-token rate, at any volume.
- $0.25 test balance. Credited at sign-up, so your first requests go through before any top-up.
- No subscription. No monthly minimum and no seat fees - you pay for the tokens you actually use.
- No identity verification. Payment needs no identity checks and no card issued in a particular country.
- All OpenAI models. The full lineup: the new GPT-5.6 family (sol, terra, luna), plus GPT-5.5, GPT-5.4 and GPT-5.4-mini.
- Streaming support. Responses stream token by token, just like the official API.
- Works with your IDE. Keep using Cursor, Copilot, Open Code and other integrations.
- OpenAI API format. Format-compatible - you only change the base URL.
- Tokens never expire. Tokens you have paid for stay with you with no expiry date.
- Pay with crypto. BTC, ETH, USDT, USDC and other popular coins.
A simple way to connect to OpenAI models
ChatGPT API Dev is a genuinely simple and affordable way to connect to OpenAI models - the API surface you already know, at half the price.
Start saving