How much does the OpenAI API cost?
Every OpenAI model we track, per 1M tokens. Batch prices are half the standard rate.
| Model | Input | Cached input | Output | Batch input | Batch output | Context |
|---|---|---|---|---|---|---|
| GPT-6 Astra | $10.00 | $1.00 | $50.00 | $5.00 | $25.00 | 1.05M |
| GPT-6.1 Sol | $2.00 | $0.10 | $10.00 | $1.00 | $5.00 | 1.05M |
| GPT-6.1 Terra | $1.00 | $0.05 | $5.00 | $0.50 | $2.50 | 1.05M |
| GPT-6 Luna | $0.10 | $0.01 | $0.50 | $0.05 | $0.25 | 1.05M |
| GPT-5.6 Solpromo to 30 Nov | $4.00 | $0.40 | $20.00 | $2.00 | $10.00 | 1.05M |
| GPT-5.6 Terra | $2.00 | $0.20 | $12.00 | $1.00 | $6.00 | 1.05M |
| GPT-5.6 Luna | $0.20 | $0.02 | $1.20 | $0.10 | $0.60 | 1.05M |
| GPT-5.5 Pro | $30.00 | Not listed | $180.00 | $15.00 | $90.00 | 1.05M |
| GPT-5.5 | $5.00 | $0.50 | $30.00 | $2.50 | $15.00 | 1.05M |
| GPT-5.4 Pro | $30.00 | Not listed | $180.00 | $15.00 | $90.00 | 1.05M |
| GPT-5.4 | $2.50 | $0.25 | $15.00 | $1.25 | $7.50 | 1.05M |
| GPT-5.4 Mini | $0.75 | $0.075 | $4.50 | $0.375 | $2.25 | 400K |
| GPT-5.2 Pro | $21.00 | Not listed | $168.00 | $10.50 | $84.00 | 400K |
| GPT-5.2 | $1.75 | $0.175 | $14.00 | $0.875 | $7.00 | 400K |
| GPT-5.1 retires 1 Apr 2027 | $1.25 | $0.125 | $10.00 | $0.625 | $5.00 | 400K |
| GPT-5 retires 11 Dec 2026 | $1.25 | $0.125 | $10.00 | $0.625 | $5.00 | 400K |
| GPT-5 Mini retires 11 Dec 2026 | $0.25 | $0.025 | $2.00 | $0.125 | $1.00 | 400K |
| GPT-5 Nano retires 11 Dec 2026 | $0.05 | $0.005 | $0.40 | $0.025 | $0.20 | 400K |
| o3-pro retires 11 Dec 2026 | $20.00 | Not listed | $80.00 | $10.00 | $40.00 | 200K |
| o3 retires 11 Dec 2026 | $2.00 | $0.50 | $8.00 | $1.00 | $4.00 | 200K |
| GPT-4.1 retires 14 Jan 2027 | $2.00 | $0.50 | $8.00 | $1.00 | $4.00 | 1.05M |
| GPT-4.1 Mini retires 14 Jan 2027 | $0.40 | $0.10 | $1.60 | $0.20 | $0.80 | 1.05M |
| GPT-4o retires 14 Jan 2027 | $2.50 | $1.25 | $10.00 | $1.25 | $5.00 | 128K |
| GPT-4o Mini retires 14 Jan 2027 | $0.15 | $0.075 | $0.60 | $0.075 | $0.30 | 128K |
From OpenAI's pricing page, checked 6 Nov 2026. See all LLM API prices for the other 6 providers.
Which OpenAI model is cheapest for a typical workload?
Of the models we track, GPT-5 Nano, at $0.75 a month for 1,000 input and 500 output tokens a request, 100 requests a day. The same traffic costs $360.00 on GPT-5.4 Pro. GPT-5 Nano shuts down on 11 Dec 2026; the cheapest model with no shutdown date is GPT-6 Luna, at $1.05 a month. Every model, cheapest first:
| Model | Per request | Per month | Per month, batch |
|---|---|---|---|
| GPT-5 Nano retires 11 Dec 2026 | $0.00025 | $0.75 | $0.375 |
| GPT-6 Luna | $0.00035 | $1.05 | $0.525 |
| GPT-4o Mini retires 14 Jan 2027 | $0.00045 | $1.35 | $0.675 |
| GPT-5.6 Luna | $0.0008 | $2.40 | $1.20 |
| GPT-4.1 Mini retires 14 Jan 2027 | $0.0012 | $3.60 | $1.80 |
| GPT-5 Mini retires 11 Dec 2026 | $0.0013 | $3.75 | $1.875 |
| GPT-5.4 Mini | $0.003 | $9.00 | $4.50 |
| GPT-6.1 Terra | $0.0035 | $10.50 | $5.25 |
| o3 retires 11 Dec 2026 | $0.006 | $18.00 | $9.00 |
| GPT-4.1 retires 14 Jan 2027 | $0.006 | $18.00 | $9.00 |
| GPT-5.1 retires 1 Apr 2027 | $0.0063 | $18.75 | $9.375 |
| GPT-5 retires 11 Dec 2026 | $0.0063 | $18.75 | $9.375 |
| GPT-6.1 Sol | $0.007 | $21.00 | $10.50 |
| GPT-4o retires 14 Jan 2027 | $0.0075 | $22.50 | $11.25 |
| GPT-5.6 Terra | $0.008 | $24.00 | $12.00 |
| GPT-5.2 | $0.0088 | $26.25 | $13.125 |
| GPT-5.4 | $0.01 | $30.00 | $15.00 |
| GPT-5.6 Solpromo to 30 Nov | $0.014 | $42.00 | $21.00 |
| GPT-5.5 | $0.02 | $60.00 | $30.00 |
| GPT-6 Astra | $0.035 | $105.00 | $52.50 |
| o3-pro retires 11 Dec 2026 | $0.06 | $180.00 | $90.00 |
| GPT-5.2 Pro | $0.105 | $315.00 | $157.50 |
| GPT-5.5 Pro | $0.12 | $360.00 | $180.00 |
| GPT-5.4 Pro | $0.12 | $360.00 | $180.00 |
How much do cached input and the Batch API save on OpenAI?
Cached input costs 10% of the input price on 13 models, 5% on GPT-6.1 Sol and GPT-6.1 Terra, 25% on o3, GPT-4.1 and GPT-4.1 Mini and 50% on GPT-4o and GPT-4o Mini. GPT-5.5 Pro, GPT-5.4 Pro, GPT-5.2 Pro and o3-pro have no cached price. A long, fixed system prompt is the easiest part of a request to serve from the cache.
On the 4 GPT-6 models, putting a prompt into the cache has its own price: 1.25 times the input rate to keep it for 30 minutes, or 2.5 times for 12 hours. On GPT-6.1 Sol that is $2.50 or $5.00 per 1M tokens written, then $0.10 per 1M each time it is read.
Batch cuts the whole bill when you can wait for results: half the input and output price on every model in the table. The example workload on GPT-6.1 Sol drops from $21.00 to $10.50 a month. Our guide to cutting OpenAI API costs measures caching and batch on a real workload.
What else changes the price of an OpenAI API call?
Three things on top of the list price. GPT-5.4 and later charge a higher rate once a prompt passes 272K input tokens; the tables here show the standard rate. Fast mode, which OpenAI called priority processing until July 2026, costs twice the standard price on the GPT-6 models. Data residency endpoints add 10% for models released on or after 5 March 2026.
Prompt size is where most estimates go wrong. OpenAI bills in tokens, not words, and our OpenAI token counter runs the o200k_base tokenizer in your browser, so the count matches the bill on GPT-5, GPT-4.1, GPT-4o and o3.
Which OpenAI models are being retired?
10 of the models in the table have shutdown dates on OpenAI's deprecation page: GPT-5, GPT-5 Mini, GPT-5 Nano, o3-pro and o3 on 11 Dec 2026; GPT-4.1, GPT-4.1 Mini, GPT-4o and GPT-4o Mini on 14 Jan 2027; GPT-5.1 on 1 Apr 2027. After that date, calls to the old model ID stop working, so switch and re-test before it, not on it.
If you are moving off GPT-5, our guide to what to move to after GPT-5 compares the replacements on price. GPT-6 Luna is the cheapest current model with no shutdown date.
Is the OpenAI API free?
Not beyond a first test. After a test request, you need at least $5 of prepaid credit to keep calling the API. The cheapest way to keep building is GPT-6 Luna at $0.10 in and $0.50 out, which runs the example workload for $1.05 a month. See is the OpenAI API free for what the trial covers, or the free LLM APIs that still have a standing free tier.
How does OpenAI API pricing compare with Claude and Gemini?
The budget, flagship and top models from each provider on the same example workload, cheapest first. OpenAI rows are highlighted. For the full price lists, see Claude API pricing and Gemini API pricing.
| Model | Provider | Input | Output | Example month |
|---|---|---|---|---|
| GPT-6 Luna | OpenAI | $0.10 | $0.50 | $1.05 |
| Claude Haiku 5.5 | Anthropic | $0.10 | $0.50 | $1.05 |
| Gemini 3.5 Flash Lite | $0.30 | $2.50 | $4.65 | |
| GPT-6.1 Sol | OpenAI | $2.00 | $10.00 | $21.00 |
| Claude Sonnet 5.5 | Anthropic | $2.00 | $10.00 | $21.00 |
| Gemini 4 Argonpromo to 30 Nov | $2.00 | $10.00 | $21.00 | |
| GPT-6 Astra | OpenAI | $10.00 | $50.00 | $105.00 |
| Claude Fable 5.1 | Anthropic | $10.00 | $50.00 | $105.00 |
Frequently asked
How much does the OpenAI API cost per 1M tokens?
From $0.05 input and $0.40 output (GPT-5 Nano) to $30.00 input and $180.00 output (GPT-5.5 Pro). GPT-6 Astra, the most expensive GPT-6 model, is $10.00 in and $50.00 out. Prices checked 6 Nov 2026.
What is the cheapest OpenAI API model?
Of the 24 OpenAI models we track, GPT-5 Nano, at $0.05 per 1M input and $0.40 per 1M output tokens. 100 requests a day of 1,000 input and 500 output tokens cost $0.75 a month on it, and half that through the Batch API. GPT-5 Nano shuts down on 11 Dec 2026; the cheapest model with no shutdown date is GPT-6 Luna, at $1.05 a month.
Is GPT-6 more expensive than GPT-5?
Not across the board. GPT-6.1 Sol is $2.00 in and $10.00 out, below GPT-5.5 at $5.00 and $30.00. GPT-6 Astra is the expensive one at $10.00 and $50.00. Compare them on your own prompt in the LLM cost calculator.
Does the OpenAI Batch API cost less?
Yes. Requests sent through the Batch API cost half the normal input and output price, in exchange for results that come back later instead of right away. The batch columns in the table above apply that 50% to every model.