MINI REASONING400K CONTEXTTEXT + VISIONPROMPT CACHINGCOMPUTER USE
GPT-5.4 mini API Pricing
OpenAI's smaller GPT-5.4 model for coding, computer use, and subagents: $0.75/M input, $4.50/M output, and $0.075/M cached input. It keeps a 400K context window and the same text-plus-image input shape as the larger GPT-5.4 family.
Input - per 1M tokens
$0.75/M
Standard model tier standard
Output - per 1M tokens
$4.50/M
Batch/Flex output is $2.25/M standard
Cached input - 90% off
$0.07/M
Prompt cache hit price -90%
Effective - agentic blend
$0.54/M
92/8 split - 82% cache
§ 01 / TERMINAL
Run the numbers.
Live calculator pre-loaded with current GPT-5.4 mini rates. Tweak cache hit rate or workload split to compare subagent, support, and computer-use workloads.
$ /mo
Workload split
Prompt cache hit rate
Tokens you can process
—
Words equivalent (English)
—
Effective rate
—
§ 02 / SCENARIOS
Real-world presets.
CHATBOT
Product support assistant
$0.005/turn
SUBAGENT
Coding helper subtask
$0.039/task
RAG
Knowledge-base support
$0.013/ticket
EXTRACTION
Invoice data extraction
$0.004/doc
COMPUTER USE
Browser workflow automation
$0.019/run
§ 03 / TOKENIZER
Paste text. See tokens. See cost.
Exact · o200k_base Auto-counts as you type
Counted with the o200k_base BPE encoding — the same tokenizer the API uses — entirely in your browser. The encoder loads on your first keystroke (~1 MB, one time); your text never leaves this page.
Characters 559
Words 85
Tokens (exact) 143 tokens
Cost as input · uncached $0.000107 USD
Cost as output · uncached $0.000643 USD
Cost as cached input $0.000011 USD
| Model | Input /M | Output /M | Effective blended | Context | Best for |
|---|---|---|---|---|---|
| GPT-5.4 mini Current | $0.75 cache $0.07 | $4.50 | $0.54 agentic 92/8 | 400K | Subagents and computer-use tasks |
| GPT-5.4 | $2.50 cache $0.25 | $15.00 | $1.80 larger model | 1.05M | Affordable OpenAI frontier work |
| GPT-5.4 nano | $0.20 cache $0.02 | $1.25 | $0.15 same blend | 400K | Classification and extraction |
| GPT-5.3-Codex | $1.75 cache $0.17 | $14.00 | $1.54 coding specialist | 400K | Agentic coding tasks |
| Gemini 2.5 Pro | $1.25 cache $0.13 | $10.00 | $1.10 tier 1 pricing | 2M | Large multimodal context |
| DeepSeek V4 Pro | $0.43 cache $0.00 | $0.87 | $0.14 promo price | 1M | Low-cost reasoning workloads |
| DeepSeek V4 Flash | $0.14 cache $0.00 | $0.28 | $0.05 same blend | 1M | Bulk low-cost traffic |
Frequently asked.
Practical GPT-5.4 mini pricing questions, with OpenAI's published rates separated from workload assumptions.
Q · 01 What is the standard GPT-5.4 mini API price? +
OpenAI lists
gpt-5.4-mini at $0.75/M input tokens, $0.075/M cached input tokens, and $4.50/M output tokens. This page stores the public list price in USD and marks the source as OpenAI's pricing docs.Q · 02 How much cheaper is GPT-5.4 mini than GPT-5.4? +
GPT-5.4 mini is 3.3x cheaper than GPT-5.4 on the shared agentic blend: about
$0.54/M versus $1.80/M. The tradeoff is model capability and context: mini has a 400K context window rather than GPT-5.4's larger 1.05M window.Q · 03 How much cheaper are Batch and Flex? +
OpenAI lists Batch and Flex for
gpt-5.4-mini at half the standard rate: $0.375/M input, $0.0375/M cached input, and $2.25/M output. That makes latency-tolerant subagent and support workloads materially cheaper.Q · 04 What does Priority processing cost? +
The Priority table lists
gpt-5.4-mini at $1.50/M input, $0.15/M cached input, and $9/M output. That is 2x the standard price, so use it only when latency matters.Q · 05 Does GPT-5.4 mini have regional pricing? +
Yes. OpenAI states that regional processing/data-residency endpoints carry a
10% uplift for gpt-5.4-mini and related GPT-5.4/5.5 models. Standard global routing uses the public standard rate.Q · 06 Does GPT-5.4 mini support computer use? +
Yes. OpenAI's model page lists computer use among the tools supported by
gpt-5.4-mini in the Responses API. GPT-5.4 nano's model page does not list computer use support, which is one practical reason to pick mini for browser or desktop automation.Q · 07 What release date is used here? +
OpenAI's model page lists the first GPT-5.4 mini snapshot as
gpt-5.4-mini-2026-03-17. This page uses 2026-03-17 as the model release date for pricing-history purposes.Q · 08 How accurate is the tokenizer estimate? +
The live widget uses an estimated
4.875 characters per token for English text and labels the tokenizer as tiktoken-o200k_base. It is good for budget planning, but exact billing can differ by language, whitespace, code, tool calls, and image inputs.