Is GPT-5 mini or DeepSeek V4 Flash cheaper?

DeepSeek V4 Flash is cheaper on most of the workload shapes tested. On a RAG workload of 12k input and 700 output tokens at 50,000 requests per month with 60% caching, GPT-5 mini costs $139.00 per month and DeepSeek V4 Flash costs $156.84. Headline rates are $0.25/$2 versus $0.44/$1.32 per million input/output tokens.

GPT-5 mini vs DeepSeek V4 Flash

DeepSeek V4 Flash is the cheaper of the two across most of the workload shapes tested below. Headline rates are $0.25 in / $2 out for GPT-5 mini versus $0.44 in / $1.32 out for DeepSeek V4 Flash, per million tokens.

Real monthly cost, three workloads

Workload	GPT-5 mini	DeepSeek V4 Flash	Cheaper
Support chatbot 1.5k in / 400 out · 100k req/mo · 30% cached	$107.38	$99.63	DeepSeek V4 Flash (1.1× cheaper)
RAG / doc Q&A 12k in / 700 out · 50k req/mo · 60% cached	$139.00	$156.84	GPT-5 mini (1.1× cheaper)
Coding agent 25k in / 2.5k out · 20k req/mo · 80% cached	$135.00	$115.60	DeepSeek V4 Flash (1.2× cheaper)

Costs include prompt caching at the stated hit rate and per-model tokenizer correction. Adjust for your own traffic →

Specifications side by side

	GPT-5 mini	DeepSeek V4 Flash
Provider	OpenAI	DeepSeek
Input / M tokens	$0.25	$0.44
Output / M tokens	$2	$1.32
Cached input	$0.025	$0.014
Context window	—	1000K
Batch discount	−50%	—
Tokenizer factor	×1	×1
Min prefix to cache	—	—
Announced retirement	—	—

Which should you pick?

On cost alone, DeepSeek V4 Flash wins the majority of the shapes above. The ranking flips depending on the input-to-output ratio, so check the shape closest to your own traffic.

Full detail: GPT-5 mini · DeepSeek V4 Flash

All figures verified 2026-08-18 against official provider documentation. Fields a provider does not publish are shown as “—” rather than estimated.