AI AI Prices

Home / Models / OpenAI / GPT-5.4

OpenAI
Fallback source
Warming up

GPT-5.4 API Pricing and Cost Calculator

Current OpenAI GPT-5.4 API prices, latest update, quick estimate, price history, similar models, and a prefilled calculator path.

Use this model pricing page for the single-model question first: current input and output price, when it was updated, whether the source is traceable, what a reference workload costs, and where to calculate your own spend.

gpt-5.4 · text · active

Calculate Cost Open compare Provider pricing page Provider changes

GPT-5.4 Calculator Card

Estimate Monthly Cost directly from this model page. Inputs update the result and the share link immediately.

Default expanded
Monthly Cost $5730.00 1,000 requests per day · 1,000 input · 500 output · 30 days
Open full calculator

Current API pricing

If you searched GPT-5.4 API pricing, start with the current stored prices below, then use Calculate Cost for workload-specific monthly spend.

Updated May 22, 22:05
Input price
$2.50 per 1M tokens
Output price
$15.00 per 1M tokens
Cached input
$0.250 per 1M tokens
Blended price
$5.63 75 / 25 input-output mix
GPT-5.4 API pricing answer GPT-5.4 currently lists $2.50 input and $15.00 output per 1M tokens. Latest stored update: May 22, 22:05. Source: PricePerToken OpenAI.
Open compare Calculate Cost Provider pricing page Provider changes

Display currency

Current prices, reference costs, recent deltas, and history values switch to the selected display currency when FX is available.

Using USD display for price cards and history values.

Pricing context and trust

This block answers the context questions around the current price: where it came from, how fresh it is, whether history is still warming up, and whether cache or batch pricing is explicitly listed.

Latest snapshot May 22, 22:05 Most recent stored public pricing point for this model.
Source PricePerToken OpenAI Source page Keep the source link visible so the latest stored number stays manually verifiable.
Source mode Fallback source Verified fallback is in use because the preferred official page is not consistently crawlable right now.
History signal Warming up Tracking is still warming up. 2 stored points across 53 days. 90d window is still awaiting depth.
Cached pricing $0.250 Cached input pricing only appears when the source exposes it clearly.
Batch discount Not listed Shown only when a provider-listed batch price is normalized. Do not mix it with realtime pricing assumptions.

Change intelligence

This strip answers the practical questions: has this model moved, how meaningful was the move, is history deep enough, and which page should you open next.

Open compare Calculate Cost
Latest move No recent delta No recent delta has been detected between the latest two stored snapshots.
Stored history 2 points · 53 days First comparison. You can compare the latest move, but longer windows are still thin and may overstate short-term noise.
Strongest window 7d change looks flat This window looks stable. Use the current price card and broader history depth to decide.
Recommended next step Open compare There is enough recent context to compare this model against alternatives rather than looking at a single price card.

Quick Estimate

2k input + 1k output, 20% cache hit, 300k requests / month

Calculate Cost
Single request estimate $0.019
Monthly estimate $5730.00
Batch discount Not listed
How to read this estimate The reference workload is a fast sanity check, not a forecast. If your prompt shape, cache ratio, or monthly traffic differs, jump to the calculator with this model preselected and adjust the assumptions there.

Similar Models

Open alternatives in compare or calculator with the same display currency before committing to this model.

Open compare
Start from this model Compare shortlist Open compare with this model selected, then add nearby candidates from the same provider or workload class. Open compare
Calculator path Prefilled workload Estimate GPT-5.4 first, then add alternatives after your token and traffic assumptions are set. Calculate Cost

Latest detected changes

Compared against the previous recorded snapshot for this model.

May 22, 22:05

You can compare the latest move, but longer windows are still thin and may overstate short-term noise.

No delta yet Waiting for another snapshot A price change needs at least two recorded points. This model currently has too little history to show a delta or nothing has changed yet.

Pricing history

Chronological snapshots captured from the active pricing source for this model.

2 points
Open glossary Data sources Export CSV

You can compare the latest move, but longer windows are still thin and may overstate short-term noise.

7d change May 10, 22:05 -> May 22, 22:05
Input 0% Output 0% Cached 0%
3 stored points across 11 days This window looks stable. Use the current price card and broader history depth to decide.
30d change Apr 19, 22:05 -> May 22, 22:05
Input 0% Output 0% Cached 0%
6 stored points across 32 days This window looks stable. Use the current price card and broader history depth to decide.
90d change Awaiting depth Awaiting depth Needs 90+ days of stored snapshots.
Tracking span Mar 30, 12:00 -> May 22, 22:05 53 stored days covered by the visible history.
Latest detected change No change yet Driven only by stored deltas, not guessed from one point.
History status First comparison You can compare the latest move, but longer windows are still thin and may overstate short-term noise.
Input Output Cached

Shared scale across visible pricing lines. Use the table below for exact values.

Captured at Input / 1M Output / 1M Cached / 1M Source
Mar 30, 12:00 $2.50 $15.00 $0.250 OpenAI
May 22, 22:05 $2.50 $15.00 $0.250 PricePerToken OpenAI

Change log

Event timeline generated from stored snapshot deltas for this model.

0 events
No event log yet Waiting for a detected change A change log entry appears only when a newer snapshot differs from the previous stored point.

Source and limits

Source PricePerToken OpenAI PricePerToken OpenAI
Recorded check Checked May 22, 22:05
Parser version pricepertoken-payload-v1
Context window 400000
Output limit 128000
Trust policy This page stays official-first whenever the provider page is crawlable. If not, verified fallback or baseline states stay explicit so the source trace never over-claims a live official crawl.
Source note Fallback snapshot from PricePerToken because OpenAI official pricing pages currently return an anti-bot challenge to server-side crawlers. Source updated at 2026-05-22T08:25:38.516428Z.

When this model detail is enough

Use this page when you need a single-model read: current price, source traceability, recent change strength, and enough stored history to judge whether the latest move is actionable.

When to escalate to compare

If a recent price cut or a stable multi-point window makes this model newly competitive, move to compare so every candidate runs against the same workload assumptions.

When to escalate to calculator

If a recent price increase changes spend assumptions or history is still thin, use calculator next so you can replace the reference workload with your own traffic, cache ratio, and budget ceiling.

GPT-5.4 API pricing FAQ

Short answers for cost, monthly estimates, user-scale planning, and cheaper alternatives.

Site FAQ
How much does GPT-5.4 cost? Current price

GPT-5.4 currently costs $2.50 per 1M input tokens and $15.00 per 1M output tokens in the latest stored snapshot.

What is the monthly cost for GPT-5.4? Quick Estimate

The built-in quick estimate uses 2k input + 1k output, 20% cache hit, 300k requests / month and returns $5730.00 per month. Use Calculate Cost to adjust token counts, request volume, cache ratio, batch mode, and budget.

Can GPT-5.4 support 100 users? Workload fit

A 100-user estimate depends on how many requests each user sends and how large the prompts and responses are. This page lists context window and output limit; the calculator link is prefilled with GPT-5.4 so you can model real traffic.

Is there a cheaper alternative to GPT-5.4? Similar Models

Use the Similar Models section above to open a shortlist in compare, then move to calculator once your workload assumptions are set. The page does not guess a winner from price alone because cached input, output mix, and batch mode can change the result.