OpenAI API Pricing (2026)
OpenAI publishes three prices per model: input, cached input, and output. This page lists the full current lineup and explains why the cached input column is the one that decides most production bills.
OpenAI model prices
List prices verified 2026-08-10. USD per 1 million tokens, standard tier.
| Model | Input $/1M | Cached input $/1M | Output $/1M | Context |
|---|---|---|---|---|
| GPT-5.6 Sol | $5.00 | $0.50 | $30.00 | 1,050,000 |
| GPT-5.6 Terra | $2.00 | $0.20 | $12.00 | – |
| GPT-5.6 Luna | $0.20 | $0.02 | $1.20 | – |
| GPT-5.5 | $5.00 | $0.50 | $30.00 | 1M |
| GPT-5 | $1.25 | $0.125 | $10.00 | 400K |
| GPT-5 mini | $0.25 | $0.025 | $2.00 | – |
| GPT-5 nano | $0.05 | $0.005 | $0.40 | – |
| GPT-4o | $2.50 | $1.25 | $10.00 | 128K |
| GPT-4o mini | $0.15 | $0.075 | $0.60 | 128K |
| GPT-4.1 | $2.00 | $0.50 | $8.00 | 1M |
GPT-5.6 Sol has a maximum output of 128K tokens.
Cached input pricing
Cached input is the rate you pay when a request repeats a prompt prefix that OpenAI has already processed. The discount is large but it is not uniform across the lineup, so read the table rather than assuming one ratio.
On the GPT-5 and GPT-5.6 families the cached rate is one tenth of the input rate. GPT-5 charges $0.125 cached against $1.25 uncached, and GPT-5.6 Sol charges $0.50 against $5.00. GPT-4.1 discounts further to one quarter, $0.50 cached against $2.00. The GPT-4o family discounts least: cached input is half price, $1.25 against $2.50 on GPT-4o.
The practical effect shows up in any workload that re-sends a stable prefix on every request, which is how agents, chat apps with system prompts, and retrieval pipelines all behave. When most input tokens bill at the cached rate, a model's effective input price is closer to its cached column than its input column. That changes rankings. GPT-4o's cached rate of $1.25 is 10x GPT-5's $0.125, a much wider gap than their uncached prices suggest. For a cache-heavy workload, compare the cached column first, then the output column, and treat the uncached input price as the rate for only the fresh part of each request.
Output prices span a wide range
Output runs from $0.40 per 1M on GPT-5 nano to $30.00 on GPT-5.6 Sol and GPT-5.5, a 75x spread. For chat workloads, where responses are long relative to prompts, the output column drives the bill more than either input column. Pick the model tier by the output work it needs to do, and let the calculator above show the crossover points for your token mix.
Related pages
- All-provider price comparison, GPT next to Claude, Gemini, DeepSeek, and Kimi.
- Claude API pricing and Gemini API pricing.
- GPT-5 context window and GPT-4o context window.
- OpenAI token counter for exact GPT token counts.