Home → Guides → OpenAI API pricing 2026: what the ChatGPT API actually costs per million tokens

OpenAI API pricing 2026: what the ChatGPT API actually costs per million tokens

Five GPT models, list prices from $2 to $10 per million input tokens. The price table is the easy part; what decides the bill is how much of your input is cached and how much each model writes back.

Price per million tokens

List prices as used for billing on this gateway, and what the same models cost here:

Model	Input	Cached input	Output	Here: input	Here: output
`gpt-6-astra`	$10.00	$1.00	$50.00	$2.00	$10.00
`gpt-5.6-sol`	$5.00	$0.50	$30.00	$1.00	$6.00
`gpt-5.6-terra`	$2.00	$0.20	$12.00	$0.400	$2.40
`gpt-6-sol`	$2.00	$0.20	$10.00	$0.400	$2.00
`gpt-6.1-sol`	$2.00	$0.10	$10.00	$0.400	$2.00

Long requests cost more. Above 272,000 input tokens the whole call is repriced: gpt-6-sol to $4 / $15, gpt-6-astra to $20 / $75. Cached input counts toward that threshold. The details are in the GPT-6 Astra pricing guide.
gpt-5.6-sol is billed here at $5 / $30. OpenAI's published list price for it has since moved to $4 / $20; this table shows the rate actually used for billing.
gpt-6.1-sol cached input has been billed at $0.10 per million — half of gpt-6-sol's $0.20 — on the 15 calls so far. A small sample; treat it as observed, not guaranteed.

ChatGPT subscription vs the API

A ChatGPT subscription and API usage are billed separately. The subscription covers the chat app; it includes no API credit. If you call a model from code, Codex, Cursor or any other tool through an API key, you pay per token, and the table above is what applies.

What a typical call costs on real traffic

All billed calls through this gateway under the 272K threshold, priced at list:

Model	Calls	Input from cache	Median output tokens	Median call at list
`gpt-5.6-sol`	699	90%	219	$0.065
`gpt-5.6-terra`	598	88%	137	$0.018
`gpt-6-astra`	397	78%	252	$0.213
`gpt-6.1-sol`	15	15%	203	$0.017

Most input is cached. 78–90% of input tokens on the three busiest models came from cache, billed at a tenth of the input rate. That is what coding agents look like: the same files and conversation re-read every turn.
Outputs are short. The median call returned 137–252 tokens. For this kind of work, the input rate and the cache rate matter more than the output rate.
gpt-6.1-sol is new here; its 15 calls are mostly uncached one-off requests, so its median cost reflects short prompts rather than agent loops.

What 1,000 typical calls cost

Using the median call above, at list price:

gpt-6-astra: about $213
gpt-5.6-sol: about $65
gpt-5.6-terra: about $18
gpt-6.1-sol: about $17

On this platform the same calls cost 20% of those figures. The spread between models is twelvefold, which is a bigger lever than any discount: routing routine work to a cheaper model usually saves more than shopping for a lower rate on the expensive one.

Which model for what

Hard reasoning, the final step a user reads: gpt-6-astra. Keep requests under 272K.
Everyday agent loops and high-volume calls: gpt-5.6-terra or gpt-6.1-sol, at roughly a twelfth of Astra's cost per call above.
When you have tuned prompts for GPT-5.6 and want to keep them: gpt-5.6-sol.

Calling the API

Any OpenAI-compatible client works. Point it at https://apiclan.us/v1 and use your key:

from openai import OpenAI
client = OpenAI(base_url="https://apiclan.us/v1", api_key="sk-your-key")
r = client.chat.completions.create(
    model="gpt-6.1-sol",
    messages=[{"role": "user", "content": "hello"}],
)
print(r.choices[0].message.content)

Top-ups are in USDT at 1 USDT for 2 credits, with no subscription and no monthly minimum. Codex setup is in the Astra guide; every other client is in the quickstart.

Start using it

No subscription, no monthly minimum, no sales call. Top up with USDT and spend what you use — 1 USDT gives you 2 credits of API balance.

Read the 30-second quickstart

Language: English · 简体中文 · 繁體中文 · 日本語 · 한국어 · Deutsch · Français · Español · Türkçe · Português · Italiano · Bahasa Indonesia · Tiếng Việt · Bahasa Melayu · Polski · Čeština · فارسی · Русский

Prices on this page are filled in automatically from live billing data. Third-party terms are quoted from that party's own documentation.