model comparison

vs

GPT 5.4 Nano vs GPT 5.4 Mini

two OpenAI models, compared on listed price, real workload cost, context and capabilities.

the short answer

  • GPT 5.4 Nano lists lower input and output prices.
  • GPT 5.4 Nano reads cached input for less ($0.02/M against GPT 5.4 Mini's $0.075/M), which decides the bill for agents that resend long context.
  • GPT 5.4 Nano retired on 8 October 2026; its numbers are kept for reference.
  • GPT 5.4 Mini retired on 8 October 2026; its numbers are kept for reference.

side-by-side specs

specGPT 5.4 NanoGPT 5.4 Mini
providerOpenAIOpenAI
input price$0.2/M$0.75/M
output price$1.25/M$4.5/M
cached input$0.02/M$0.075/M
blended price (3:1)$0.463/M$1.688/M
context window400k tokens400k tokens
max output128k tokens128k tokens
inputtext, imagestext, images
structured outputyesyes
weightsclosedclosed
releasedMarch 17, 2026March 2026
availabilityretired 8 October 2026retired 8 October 2026

same work, both models

workloadGPT 5.4 NanoGPT 5.4 Mini
one chat reply3,000 tokens in, 500 outGPT 5.4 Mini costs 3.7× as much as GPT 5.4 Nano$0.0012$0.0045
summarise a 100-page report50,000 tokens in, 1,500 out, one request$0.01$0.04
1,000 classification calls500 tokens in, 20 out, per call$0.13$0.47
coding-agent session2M tokens in across ~60 turns, 60,000 out, no cache$0.48$1.77
same session, cached85% of input read from the prompt cache, the rest written to it$0.17$0.62

at listed rates. token counts are typical sizes, not measurements.

when to pick which

GPT 5.4 Nano

GPT 5.4 Nano was the smallest, cheapest tier of OpenAI's GPT-5.4 family, below GPT 5.4 Mini, aimed at simple high-volume tasks. it retired on 2026-10-08, replaced by the cheaper GPT-6 Luna.

pick it if

  • you ran classification, extraction, ranking or simple sub-agent work
  • you needed the cheapest GPT-5.4 tier with image input
  • you reused long prompts, since cached reads billed at a tenth of input

skip it if

  • GPT-6 Luna is available: half the input price and a 1M window
  • you needed tool search or computer use, which Nano lacked

GPT 5.4 Mini

GPT 5.4 Mini was the larger of OpenAI's two small GPT-5.4 tiers, above Nano, aimed at coding, computer use and subagents. it retired on 2026-10-08, replaced by the cheaper GPT-6 Luna.

pick it if

  • you ran coding or subagent loops that needed more than Nano
  • you reused long prompts, since cached reads billed at a tenth of input
  • you wanted reasoning effort adjustable from none up to xhigh

skip it if

  • GPT-6 Luna is available: larger window at a fraction of the price

faq

which is cheaper, GPT 5.4 Nano or GPT 5.4 Mini?

GPT 5.4 Nano lists lower input and output prices. on a coding-agent session (2M tokens in, 60,000 out) GPT 5.4 Nano costs $0.48 and GPT 5.4 Mini $1.77.

try either model

switch between models inside the same octomind session.

prices from the catalog of 2026-11-18