As of November 9, 2026, OpenAI’s pricing documentation lists GPT-5 (gpt-5) at $1.25 per 1 million input tokens, $0.125 per 1 million cached input tokens, and $10 per 1 million output tokens under its Standard pricing table. The model page marks GPT-5 as deprecated. These are API token prices, not a ChatGPT subscription fee. OpenAI displays the rates with “$”; its cited pricing table does not spell out an ISO currency code, so confirm the currency and billing terms shown for your account before purchasing. (developers.openai.com)
GPT-5 API rates
OpenAI’s pricing documentation separates Standard rates from Batch and other processing options. The table below uses the listed Standard rates, in dollars per 1 million tokens—not the separately listed Batch rates. The displayed dollar sign is retained as OpenAI presents it; the pricing table excerpt does not specify a currency code. (developers.openai.com)
| API model | Standard input | Cached input | Standard output | Model-page status |
|---|---|---|---|---|
GPT-5 (gpt-5) |
$1.25 | $0.125 | $10.00 | Deprecated |
GPT-5 mini (gpt-5-mini) |
$0.25 | $0.025 | $2.00 | Deprecated |
GPT-5 nano (gpt-5-nano) |
$0.05 | $0.005 | $0.40 | Deprecated |
Rates are per 1 million tokens. The Standard-rate figures and cached-input prices appear in OpenAI’s pricing documentation; the model pages identify the legacy models as deprecated. Deprecated does not by itself establish that API access has ended. Check the model list available to your API project, and do not assume a shutdown date unless OpenAI documents one. (developers.openai.com)
API charges are separate from ChatGPT subscription billing. Your API cost depends on the model and usage, and can include applicable tool charges in addition to token charges. (help.openai.com)
Estimate a production API bill
For Standard usage without cached input, estimate token charges as:
Cost = (input tokens ÷ 1,000,000 × input rate) + (output tokens ÷ 1,000,000 × output rate)
For a hypothetical month of 1,000 requests, each averaging 10,000 input tokens and 1,000 billable output tokens:
- Input: 1,000 × 10,000 = 10 million tokens; 10 × $1.25 = $12.50.
- Output: 1,000 × 1,000 = 1 million tokens; 1 × $10 = $10.00.
- Illustrative total: $22.50, using the listed GPT-5 Standard rates above.
This is an arithmetic example, not a forecast. The output assumption means all billable output tokens, not just the visible answer: OpenAI says internal reasoning tokens count toward output usage and are billed at the output rate. Use the API response’s usage totals when estimating actual costs; for a complete request, the reported counts can include more than plain-text visible content. (help.openai.com)
If usage reporting shows input tokens billed at the cached-input rate, calculate those separately. For example, if 8 million of the 10 million monthly input tokens in this scenario are actually billed as cached and 2 million at the regular input rate, the input charge is (8 × $0.125) + (2 × $1.25) = $3.50. Add the $10 output charge for an illustrative total of $13.50. This is conditional arithmetic, not a guarantee of cache hits or savings. Prompt caching is enabled by default on supported models, but a matching reusable prefix is required; a repeated session alone does not guarantee a cache hit. (developers.openai.com)
Which option should you budget for?
For an existing integration, start with the exact model ID and rate used by your workload. OpenAI marks GPT-5, GPT-5 mini, and GPT-5 nano as deprecated, but that label alone does not tell you whether a particular project can still use them. Confirm access in your API model list and plan for migration rather than treating a deprecated model as a safe default for a new system. The cited model information does not establish a shutdown date. (developers.openai.com)
For a new workload, OpenAI documents these GPT-5.6 options in its pricing and model materials:
| Option | API model ID | Standard input / cached input / output per 1M tokens | Documented positioning |
|---|---|---|---|
| GPT-5.6 Terra | gpt-5.6-terra |
$2.00 / $0.20 / $12.00 | Balanced model for everyday work |
| GPT-5.6 Luna | gpt-5.6-luna |
$0.20 / $0.02 / $1.20 | Cost-sensitive, high-volume workloads |
OpenAI describes the GPT-5.6 family as generally available. However, a rate-limit entry alone does not confirm access to a specific model; the Luna model page advises checking the Playground or API model list if the model is missing. Treat Terra and Luna as budget candidates, not guaranteed fits: test representative requests for quality, latency, and actual token usage before choosing. (developers.openai.com)
Do not substitute GPT-5.4 or GPT-5.5 rates for GPT-5. Before committing a production budget, verify the precise model ID, processing tier, current rates, currency, account access, and any tool charges in OpenAI’s pricing and model documentation. (developers.openai.com)
Sources
GPT-5.6 Luna Model — OpenAI API
Understanding and counting tokens — OpenAI Help Center
Managing billing for ChatGPT and the API platform — OpenAI Help Center
GPT-5.6: Frontier intelligence that scales with your ambition