GPT-6 API Pricing per Million Tokens for Developers in Pakistan
GPT-6 API prices per million tokens: Astra $10 in and $50 out, Sol $2 and $10, Luna $0.10 and $0.50. See Batch, Fast and a Rs 97 chatbot cost example. Read now.
OpenAI's API charges per million tokens at Standard speed: GPT-6 Astra $10 input and $50 output, GPT-6.1 Sol $2 and $10, and GPT-6 Luna $0.10 and $0.50. Batch jobs cost half, Fast costs double and Ultrafast six times, according to OpenAI's pricing page.
A developer in Pakistan building a chatbot, a tool for a software house or a student project pays for the API separately from any ChatGPT plan. We read OpenAI's API pricing page on 22 October 2026. A token is a small chunk of text, roughly three quarters of an English word. All prices below are in US dollars per million tokens, so the rupee figures are only a guide.
Standard prices run from $0.10 to $50 per million tokens
At Standard speed OpenAI lists three GPT-6 models. Astra is the strongest and costs the most. GPT-6.1 Sol costs one fifth of Astra's price. Luna is the cheapest. Cached input, tokens the API has seen before, costs a tenth of normal input.
| Model | Input | Cached input | Output |
|---|---|---|---|
| gpt-6-astra | $10.00 | $1.00 | $50.00 |
| gpt-6.1-sol | $2.00 | $0.10 | $10.00 |
| gpt-6-luna | $0.10 | $0.01 | $0.50 |
The page also lists cache write prices, for holding a prompt in memory for 30 minutes or 12 hours. For Astra these are $12.50 and $25. The page has a switch between short and long context. We read the default view, so check the switch before you budget a long-document job.
Batch halves the price, Fast doubles it and Ultrafast multiplies it by six
OpenAI's Batch prices are exactly half of Standard for these models: Astra $5 and $25, Sol $1 and $5, Luna $0.05 and $0.25. Fast is double, and Ultrafast is six times for Astra and Sol. Luna has no Ultrafast price on the page.
| Mode | Astra in / out | Sol in / out | Luna in / out |
|---|---|---|---|
| Standard | $10 / $50 | $2 / $10 | $0.10 / $0.50 |
| Batch | $5 / $25 | $1 / $5 | $0.05 / $0.25 |
| Fast | $20 / $100 | $4 / $20 | $0.20 / $1 |
| Ultrafast | $60 / $300 | $12 / $60 | Not listed |
Batch is built for work that can wait, such as tagging 10,000 product descriptions overnight. OpenAI also notes that regional processing for data residency adds a 10% uplift for models released on or after 5 March 2026. Fast mode was called Priority processing until 30 July 2026.
A thousand chatbot replies cost about Rs 97 on Luna and Rs 9,700 on Astra
Say your bot handles 1,000 questions, each with 1,000 input tokens and 500 output tokens. That is 1 million in and 0.5 million out. Luna costs $0.35, Sol $7 and Astra $35. At Rs 277 to the dollar, that is about Rs 97, Rs 1,939 and Rs 9,695.
These figures are our arithmetic from the Standard table and ignore caching. The gap is why most apps start on the cheap model and call the stronger one only for hard cases. Compare with other vendors in the Claude Haiku 5.5 API price.
The API is billed apart from ChatGPT plans
OpenAI states that API use is not included in Plus or Go and is billed independently. Subscription usage in Work and Codex is separate too, and token prices should not be used to estimate it. A developer therefore pays twice if they want both.
Codex users can also run extra local chats with an API key at API rates when a plan limit is reached, per OpenAI. The credit side of the same models is in credits for Fast and Ultrafast, and plan prices are in the plan comparison.
Plan a budget before you ship
Test on Luna first, measure real tokens, then price the stronger model for the few calls that need it. Watch for hidden growth: chat history is re-sent with each turn, so long conversations cost more per message than short ones.
- Create an API account and note the billing currency and method on the page.
- Run 50 real requests on Luna and record input and output tokens.
- Multiply by the Standard prices above to get a cost per request.
- Move heavy tasks to Batch if they can wait.
- Set a spending limit before launch.
Paying for the API from Pakistan needs a card that allows foreign online payments. Read tax on card payments abroad and how freelancers get paid in Pakistan to plan the dollars in and out.
Common questions about GPT-6 API pricing
How much does the GPT-6 API cost?
At Standard speed, Astra is $10 input and $50 output, Sol $2 and $10, and Luna $0.10 and $0.50 per million tokens.
Is the API included in ChatGPT Plus?
No. OpenAI says API usage is not included and is billed independently.
What is the cheapest GPT-6 model?
GPT-6 Luna, at $0.10 per million input tokens and $0.50 per million output tokens.
What does Batch mean?
Batch is for jobs that can wait. OpenAI prices it at half the Standard rate on these models.
Are there extra charges for data residency?
OpenAI says regional processing endpoints carry a 10% uplift for models released on or after 5 March 2026.
Can I pay for the API in rupees?
We did not find a rupee price for the API on the page we read. Check the billing page in your account.
How we verified this
What we checked, where we read it, and what we could not confirm.
Last checked 22 October 2026. We read OpenAI's API pricing page and its Work and Codex pricing page. We read the default Standard view and did not read the long-context view. Sample costs and rupee figures are our arithmetic, using the State Bank of Pakistan's 22 October rate of about 277. We did not call the API.
About the author
Technology Journalist & Multimedia Producer
Ahmad Ali is a technology journalist and multimedia producer at Pakistan Era, joining the team in October 2026. He covers consumer technology and digital life for Pakistani readers, combining research with hands-on product testing.
