OpenAI GPT-4o API pricing
As of November 12, 2026, OpenAI’s GPT-4o model page lists $2.50 per 1 million standard input tokens and $10 per 1 million output tokens. It lists cached input at $1.25 per 1 million tokens. These are API token rates in USD—not a ChatGPT subscription price. The model page identifies the API model as gpt-4o; it also lists dated snapshots separately, so don’t assume the rates or availability apply identically to every snapshot. (developers.openai.com)
Availability matters: OpenAI’s model page labels GPT-4o deprecated, while listing the gpt-4o alias and its dated snapshots with individual statuses. Confirm the model is available to your API account before planning a new production integration around it. (developers.openai.com)
| GPT-4o API pricing dimension | Listed price per 1M tokens | Applies to |
|---|---|---|
| Standard input | $2.50 | Synchronous API usage, uncached input |
| Cached input | $1.25 | Input tokens that qualify for prompt caching |
| Output | $10.00 | Standard output tokens |
| Batch input | $1.25 | Eligible requests submitted through the Batch API |
| Batch output | $5.00 | Eligible requests submitted through the Batch API |
OpenAI says Batch API usage costs 50% less than synchronous API usage. The Batch rates above are calculated from that discount and GPT-4o’s listed standard rates. Batch is asynchronous, with results returned within 24 hours; use it only when that turnaround works for the workload. (developers.openai.com)
How to estimate your GPT-4o API cost
For standard, uncached usage, calculate input and output separately:
Estimated token cost = (input tokens ÷ 1,000,000 × $2.50) + (output tokens ÷ 1,000,000 × $10.00)
For a request using 2,000 input tokens and 500 output tokens:
- Input: 2,000 ÷ 1,000,000 × $2.50 = $0.005
- Output: 500 ÷ 1,000,000 × $10.00 = $0.005
- Total: $0.01 per request
At 1,000 requests a day with those same token counts, that is $10 per day, or $300 for a 30-day month. This is a worked estimate using standard, uncached rates—not a prediction of your bill. Actual token counts and eligible pricing dimensions can vary by request.
Another standard-rate example: 100,000 input tokens and 25,000 output tokens cost $0.25 for input plus $0.25 for output, for a $0.50 token subtotal.
Cached input and Batch pricing
Prompt caching is enabled automatically for supported models, but a cache discount is not guaranteed just because a prompt is reused. OpenAI describes caching as matching an exact, reused prompt prefix; its published GPT-4o caching guidance says eligible prompts must be longer than 1,024 tokens, with cached portions matched in 128-token increments. The listed cached-input rate is $1.25 per 1 million tokens. (developers.openai.com)
For workloads that can run asynchronously, the Batch API offers a 50% discount versus synchronous API pricing and a 24-hour completion window. At half the listed standard GPT-4o rates, that works out to $1.25 per 1 million input tokens and $5 per 1 million output tokens. OpenAI notes that Batch is not supported for every model; check the model reference and your account’s access before relying on it. Don’t assume cached-input and Batch discounts stack—the cited pricing guidance does not establish combined eligibility or rates. (developers.openai.com)
The estimates above cover GPT-4o token usage only. Separately billed tools or services may add charges. OpenAI notes that some tool-specific models have a fee per tool call. (developers.openai.com)
GPT-4o API price versus ChatGPT price
These figures are API usage rates, based on API token usage. They are not monthly ChatGPT plan prices. OpenAI’s Help Center says GPT-4o was retired from ChatGPT on February 13, 2026, while its API model page labels GPT-4o deprecated. Check current API model access before using these rates as the basis for a new deployment. (developers.openai.com)
Bottom line
For standard GPT-4o API usage, budget $2.50 per million input tokens and $10 per million output tokens; eligible cached input is listed at $1.25 per million. Batch API usage is 50% cheaper than synchronous usage, with a 24-hour turnaround, so the corresponding rates are $1.25 per million input tokens and $5 per million output tokens. Calculate input and output separately, and confirm model access and any discount eligibility before committing to a production budget. (developers.openai.com)