How much does the GPT-5.4 mini API cost?

GPT-5.4 mini lists at $0.75 per million input tokens and $4.5 per million output tokens as of August 2026. On a realistic task of 1,500 input and 500 output tokens across 3 calls, that is $0.01013 per task, or $10,125 per million tasks.

Is GPT-5.4 mini expensive compared to other models?

GPT-5.4 mini ranks #5 of 10 tracked models on cost per task. It costs 6.43x more than the cheapest tracked model (Mistral Small), while the most expensive (GPT-5.5) costs 6.67x more than it.

When does self-hosting beat the GPT-5.4 mini API?

Against a $2/hour dedicated GPU ($1,460/month), self-hosting becomes cheaper than GPT-5.4 mini above 144,198 tasks per month, within that box's physical capacity of 788,400 tasks.

Model pricing · OpenAI · mid tier · August 2026

GPT-5.4 mini API pricing, in numbers that matter

List price, real cost per task, rank against the field, and the self-hosting break-even, all computed from our open dataset.

Input, $/1M tokens

$0.75

Output, $/1M tokens

$4.5

Cost per task

$0.01013

Cost rank

#5 of 10

The task profile behind every figure: 1,500 input and 500 output tokens per call, 3 calls per task. On that profile GPT-5.4 mini costs $0.01013 per task and $10,125 per million tasks. Output is billed at 6x the input rate, so despite the profile sending three times more input than output, output makes up 67% of the bill. Capping response length is the highest-return prompt-level optimisation on this model.

GPT-5.4 mini against the field

#	Model	$/task	vs GPT-5.4 mini	Head-to-head
1	Mistral Small	$0.00158	0.16x	compare
2	DeepSeek-V4-Flash	$0.00396	0.39x	compare
3	Gemini 2.5 Flash	$0.00510	0.50x	compare
4	Grok 4.3	$0.00938	0.93x	compare
5	GPT-5.4 mini	$0.01013	-	-
6	Claude Haiku 4.5	$0.01200	1.19x	compare
7	Claude Sonnet 5	$0.02400	2.37x	compare
8	Gemini 3.1 Pro	$0.02700	2.67x	compare
9	Claude Opus 4.8	$0.06000	5.93x	compare
10	GPT-5.5	$0.06750	6.67x	compare

Self-hosting break-even

Against a reference $2/hour dedicated instance ($1,460/month fixed, 788,400 tasks/month capacity at full utilisation), the break-even against GPT-5.4 mini is 144,198 tasks per month. That sits inside the box's capacity, so past that volume self-hosting genuinely wins on cost. Work through your own numbers in the $/task calculator or the break-even guide.

Frequently asked

How much does the GPT-5.4 mini API cost?: GPT-5.4 mini lists at $0.75 per million input tokens and $4.5 per million output tokens as of August 2026. On a realistic task of 1,500 input and 500 output tokens across 3 calls, that is $0.01013 per task, or $10,125 per million tasks.
Is GPT-5.4 mini expensive compared to other models?: GPT-5.4 mini ranks #5 of 10 tracked models on cost per task. It costs 6.43x more than the cheapest tracked model (Mistral Small), while the most expensive (GPT-5.5) costs 6.67x more than it.
When does self-hosting beat the GPT-5.4 mini API?: Against a $2/hour dedicated GPU ($1,460/month), self-hosting becomes cheaper than GPT-5.4 mini above 144,198 tasks per month, within that box's physical capacity of 788,400 tasks.