OpenAI API pricing
Every OpenAI API model's price per million tokens, copied from OpenAI's own pricing page on 20 November 2026: input, cached input, output and batch, plus the rules that change the bill. Current models range from $0.1 to $8 per million input tokens.
OpenAI API list prices checked 20 November 2026, per million tokens (input/output): GPT-6.2 Astra $8/$40; GPT-6.2 Sol $2/$10; GPT-6.1 Terra $1/$5; GPT-6 Luna $0.1/$0.5. Batch is half price.
OpenAI API price per million tokens
Source: OpenAI's pricing page, checked 20 November 2026. AiGanak re-checks every price weekly.
| Model | Input | Cached input | Cache write | Output | Batch in / out |
|---|---|---|---|---|---|
| GPT-6.2 Astra Flagship · Released 17 Nov 2026 | $8 | $0.4 | $10 | $40 | $4 / $20 |
| GPT-6.2 Sol Mid · Released 17 Nov 2026 | $2 | $0.1 | $2.50 | $10 | $1 / $5 |
| GPT-6.1 Terra Mid · Released 26 Oct 2026 | $1 | $0.05 | $1.25 | $5 | $0.5 / $2.50 |
| GPT-6 Luna Small | $0.1 | $0.01 | $0.125 | $0.5 | $0.05 / $0.25 |
| GPT-6 Astra Flagship · Previous version | $10 | $1 | $12.50 | $50 | $5 / $25 |
| GPT-6.1 Sol Mid · Previous version | $2 | $0.1 | $2.50 | $10 | $1 / $5 |
Rules that change the bill
- Prompt caching: a 30-minute cache write costs 1.25 times the input rate and a 12-hour write 2.5 times. Reading from cache costs 5% of the input rate on GPT-6.2 Astra, GPT-6.2 Sol and GPT-6.1 Terra, and 10% on GPT-6 Astra and GPT-6 Luna.
- Batch and Flex: 50% of standard rates.
- Long prompts: requests with more than 272K input tokens pay 2 times the input and cache rates and 1.5 times the output rate for the whole request, on every current GPT-6 model.
- Fast mode costs twice the standard rate. Ultrafast costs 3 times the standard rate through 28 February 2027 and 6 times from 1 March 2027.
- Reasoning tokens are billed as output, so a model that thinks at length costs more than its rates suggest.
Which GPT model is cheapest per request
A support answer: 6,000 input tokens (4,000 cached instructions and help text), 500-token answer. Standard list prices. Cheaper is only better if quality holds on your own examples.
| Model | Per answer | Per 1,000 | vs cheapest |
|---|---|---|---|
| GPT-6 Luna | $0.00049 | $0.490 | cheapest |
| GPT-6.1 Terra | $0.00470 | $4.70 | 9.6× |
| GPT-6.2 Sol | $0.00940 | $9.40 | 19.2× |
| GPT-6.2 Astra | $0.038 | $37.60 | 76.7× |
OpenAI vs Anthropic and Google, tier by tier
The cheapest current model in each tier for the same support answer.
| Tier | Anthropic | OpenAI | |
|---|---|---|---|
| Flagship | Claude Opus 5.5 $0.019 per answer | GPT-6.2 Astra $0.038 per answer | Gemini 4 Carbon $0.00940 per answer |
| Mid | Claude Sonnet 5.5 $0.00940 per answer | GPT-6.1 Terra $0.00470 per answer | Gemini 3.8 Flash $0.00367 per answer |
| Small | Claude Haiku 5.5 $0.00049 per answer | GPT-6 Luna $0.00049 per answer | Gemini 3.1 Flash-Lite $0.00135 per answer |
Full price lists: Anthropic Claude API pricing · Google Gemini API pricing. For a side-by-side of two providers, read OpenAI vs Anthropic API pricing.
Head-to-head: GPT-6 Astra vs Claude Opus 5.5 · GPT-6 Astra vs Claude Fable 5.2 · GPT-6 Astra vs Gemini 4 Argon · GPT-6.1 Sol vs Claude Sonnet 5.5 · GPT-6.1 Sol vs Gemini 3.8 Flash · GPT-6.1 Terra vs Claude Sonnet 5.5 · GPT-6.1 Terra vs Gemini 3.8 Flash · GPT-6.1 Sol vs GPT-6.1 Terra · GPT-6 Luna vs Claude Haiku 5.5 · GPT-6 Luna vs Gemini 3.5 Flash-Lite · GPT-6.2 Astra vs Claude Fable 5.2 · GPT-6.2 Astra vs GPT-6 Astra · GPT-6.2 Astra vs Gemini 4 Carbon · GPT-6.2 Sol vs Claude Sonnet 5.5
Cutting the bill rather than comparing rates? Start with the guides to prompt caching and the Batch API, or see what you spend by model with the free LLM Bill Analyzer.
OpenAI API pricing questions
Can't find it? Press ⌘K and ask.
How much does the OpenAI API cost per million tokens?
At OpenAI's standard list prices checked 20 November 2026: GPT-6.2 Astra $8 input and $40 output; GPT-6.2 Sol $2 input and $10 output; GPT-6.1 Terra $1 input and $5 output; GPT-6 Luna $0.1 input and $0.5 output. Cached input and batch requests cost less; see the table on this page.
What is the cheapest GPT model?
GPT-6 Luna is the cheapest current OpenAI model for a typical support answer (6,000 input tokens, 4,000 of them cached, and a 500-token reply) at $0.00049 per request, against $0.038 on GPT-6.2 Astra. Check answer quality on your own examples before choosing on price.
Does OpenAI offer a batch discount?
Yes. OpenAI charges 50% of standard rates for batch requests, which return within 24 hours. For example, GPT-6.2 Astra costs $4 input and $20 output per million tokens in batch.
Is OpenAI raising its API prices?
No price increase was listed on OpenAI's pricing page when AiGanak checked it on 20 November 2026. AiGanak re-checks prices weekly and emails alerts when one changes.
Get an email when a OpenAI API price changes
OpenAI, Anthropic and Google change API prices with little notice. We check their pricing pages every week and email you the old price, the new price and the date. Free, a few emails a month at most.
Paying OpenAI more than you need to?
The LLM Cost Report prices a sample of your real requests on every current model and language and sends a ranked savings plan with code, in two working days.