Understanding OpenAI API Pricing: An Overview
OpenAI offers a range of powerful APIs, each with its own pricing structure based on factors like model, input/output tokens, and processing type. This guide breaks down the costs associated with different OpenAI models and services, helping you make informed decisions for your AI development needs.
Pricing is primarily determined by token usage (input and output), with variations for cached inputs, fine-tuning, and specialized APIs like the Realtime API and Image Generation API. Batch API offers cost saving for asynchronous task processing.
Flagship GPT-5 Pricing Details
The GPT-5 family offers varying levels of performance and cost. GPT-5 is the top-tier model, ideal for complex tasks. GPT-5 mini provides a faster, more affordable option for well-defined tasks. GPT-5 nano is the most efficient for summarization and classification.
GPT-5 costs $1.250 per 1M input tokens, $0.125 per 1M cached input tokens, and $10.000 per 1M output tokens. GPT-5 mini is priced at $0.250/$0.025/$2.000 per 1M tokens for input/cached input/output, respectively. GPT-5 nano offers the lowest cost at $0.050/$0.005/$0.400 per 1M tokens.
Fine-Tuning Customizing Models with
Fine-tuning allows you to customize OpenAI models for specific use cases, potentially improving performance. This section details the pricing for fine-tuning GPT-4.1 and o4-mini models.
GPT-4.1 fine-tuning costs $3.00/$0.75/$12.00 per 1M tokens for input/cached input/output, with a training cost of $25.00 per 1M tokens. o4-mini reinforcement fine-tuning is priced at $4.00/$1.00/$16.00 per 1M tokens for input/cached input/output, and training costs $100.00 per training hour.
“Experiment with different models in the Playground to explore which provides the best price-performance trade-off for your usage.
OpenAI Recommendation
Explore OpenAI Pricing Tools
Calculate your estimated costs with these helpful resources.
OpenAI Pricing Calculator
Estimate the cost of image generation based on resolution and tile size.
Usage Tracking Dashboard
Monitor your token usage and spending in real-time.
Realtime API Pricing Breakdown
The Realtime API enables low-latency, multimodal experiences. Pricing varies based on the model (gpt-realtime, GPT-4o mini) and modality (text, audio, image).
For example, gpt-realtime (text) costs $4.00/$0.40/$16.00 per 1M tokens, while gpt-realtime (audio) costs $32.00/$0.40/$64.00 per 1M tokens for input/cached input/output. GPT-4o mini (text) costs $0.60/$0.30/$2.40 per 1M tokens and (audio) costs $10.00/$0.30/$20.00 per 1M tokens.
The Image Generation API utilizes GPT-image-1 and costs $5.00/$1.25 and $10.00/$2.50 per 1M input/cached input tokens, and $40.00 per 1M output tokens for the latter. Image outputs range from $0.01 to $0.17 based on quality and size.
Tools Assistants API and Built-in Pricing
The Assistants API and Chat Completions API do not have separate pricing. Tokens are billed at the rate of the underlying language model.
Built-in tools like Code Interpreter cost $0.03 per session, File Search Storage costs $0.10 per GB per day (first GB free), and File Search Tool Calls (Responses API only) cost $2.50 per 1k calls. Web search tool calls vary in price depending on the model.