OpenAI GPT-5 API Pricing: Complete Cost Guide 2026
Full pricing breakdown for every current GPT-5 model — GPT-5.5, GPT-5.4, GPT-5.4 Mini, and GPT-5.4 Nano — with monthly cost estimates and a comparison to Claude and Gemini.
GPT-5 model tiers and pricing
OpenAI's GPT-5 family spans five tiers in 2026. GPT-5.5 is the current frontier at $5.00 per million input tokens and $30.00 per million output tokens, with a 272K context window. GPT-5.5 Pro ($30.00/$180.00) targets enterprise workloads requiring maximum capability and a one-million token context.
GPT-5.4 ($2.50/$15.00) covers standard production use cases at 272K context. GPT-5.4 Mini ($0.75/$4.50) is the sweet spot for most teams — strong enough for chat, coding, and classification at a fraction of the flagship cost. GPT-5.4 Nano ($0.20/$1.25) and GPT-5 Nano ($0.05/$0.40) target high-volume, lower-complexity tasks like routing, tagging, and extraction.
Monthly cost estimates at common usage levels
At 10M tokens per month (70% input, 30% output), GPT-5.4 Mini costs around $6.60/month — suitable for a small production chatbot or coding tool. GPT-5.4 at the same volume runs about $22/month. GPT-5.5 costs $44/month for that workload.
Heavy agentic applications that process 500M tokens per month see a much larger spread: GPT-5.4 Nano at $132/month versus GPT-5.4 at $1,100/month. For output-heavy tasks such as code generation or report drafting, the 30% output assumption understates the true cost — recalculate with your actual input-to-output ratio using the calculator.
How GPT-5 compares to Claude and Gemini
GPT-5.4 ($2.50/$15.00) sits in the same input price tier as Gemini 3.5 Flash ($1.50/$9.00) and just above Claude Sonnet 4.6 ($3.00/$15.00). On output price, all three are similar. Gemini 3.5 Flash is the cheapest of the three for output-heavy tasks; GPT-5.4 and Claude Sonnet are nearly identical.
For budget workloads, GPT-5.4 Nano ($0.20/$1.25) and GPT-5 Nano ($0.05/$0.40) significantly undercut Claude Haiku 4.5 ($1.00/$5.00) and Gemini 3.1 Flash-Lite ($0.25/$1.50). If task quality allows, switching to OpenAI's nano tiers can cut costs by 80 to 95 percent versus mid-tier alternatives.
Choosing the right GPT-5 tier
Start with GPT-5.4 Mini as the default for new products. It handles the majority of production tasks — chat, code review, document processing, and customer support — at a price that keeps margins healthy. Move up to GPT-5.4 or GPT-5.5 only when task complexity requires it and the quality difference is measurable.
GPT-5.4 Nano and GPT-5 Nano work best as pre-processing layers: routing incoming requests, filtering irrelevant inputs, generating simple structured outputs, and running low-stakes classification. Combining a nano model for triage with a mid-tier model for synthesis often cuts total token costs by 40 to 60 percent without degrading user-facing quality.
Estimate your own workload
Use the calculator to compare your expected API bill with a purchased or rented GPU setup.
Open calculator