GPT and ChatGPT Pricing in 2026: Plans and API Costs
By Adam, AI Engineer · 1 November 2026 · 4 min read
ChatGPT costs $0 on Free and $20 a month on Plus, with Pro plans at $100, $200 and $500. The GPT API is billed separately per token, from $0.10 per million input tokens on GPT-6 Luna to $10 on GPT-6 Astra. These are OpenAI's published US prices on 1 November 2026, before any tax.
ChatGPT plans for people
| Plan | Price per month | Good for |
|---|---|---|
| Free | $0 | Trying it out |
| Go | Launched at $8 in the US; varies by country | Light regular use on a small budget |
| Plus | $20 | Most individual users |
| Pro | $100, $200 or $500 | Heavy daily use |
Pro is billed monthly with no annual option. The Go launch post notes that local prices vary, so check the pricing page for your country.
ChatGPT plans for teams
ChatGPT Business has two seat types. A Standard seat costs $25 a month, or $20 billed yearly. A Premium seat costs $125 a month, or $100 billed yearly, with higher usage for ChatGPT Work and Codex.
GPT API prices
These are per million tokens from OpenAI's API pricing page. The last column uses one example: 100,000 requests a month, each with 3,000 input tokens and 400 output tokens, and no caching.
| Model | Input | Cached input | Output | Example month |
|---|---|---|---|---|
| GPT-6 Luna | $0.10 | $0.01 | $0.50 | $50 |
| GPT-6.1 Terra | $1 | $0.05 | $5 | $500 |
| GPT-6.1 Sol | $2 | $0.10 | $10 | $1,000 |
| GPT-6 Astra | $10 | $1 | $50 | $5,000 |
How to pay less
- Cached input: on GPT-6.1 Terra, serving 70% of the input from cache takes the example from $500 to about $301, before cache write fees. The model page lists cache writes at $1.25 per million tokens.
- Batch processing: delayed jobs cost half, so the Terra example falls to $250.
- Move older features to new models: GPT-5.6 Terra costs $2 input and $12 output, so the same example costs $1,080 on it. Test the newer model first, then switch if quality holds.
Costs that are easy to miss
- Long prompts: on GPT-6.1 Terra, prompts over 272,000 input tokens cost 2x for input and cache, and 1.5x for output, across the full request.
- Web search: $10 per 1,000 calls, plus search content tokens at the model's rate.
- Separate billing: a ChatGPT subscription and API usage are different products, so a Plus plan does not pay for your app's API calls.
Which option should you choose?
- You just use it yourself: start with Free, then Plus if you hit limits. Move to Pro only if you use it heavily every day.
- Your team uses it every day: Business gives you one managed workspace, and you can mix seat types.
- You are building a product: use the API, and start your tests on GPT-6.1 Terra or GPT-6 Luna.
Compare GPT with Claude and Gemini on your own numbers in the free AI cost calculator, or read our Claude pricing and Gemini pricing guides.