Guides / gpt-5.4 pro pricing

GPT-5.4 Pro pricing, and the missing cached rate

GPT-5.4 Pro pricing is $30.00 per million input tokens and $180.00 per million output tokens, and GPT-5.5 Pro bills identically. Neither carries a cached rate, so a repeated system prompt pays full price every call. Culpa, a local-first LLM cost, margin, and forecast ledger, meters the repeated share of your input so that cost stops being invisible.

Why this happens

Two things make the Pro tier behave differently from everything else on the roster. It's twelve times GPT-5.4 on input and twelve times on output, so the usual intuitions about rounding errors stop applying. And this price book records no cached input rate for either Pro model, which removes the one lever that normally rescues a long repeated prompt. On Pro, prompt length is paid in full, forever.

What this usually looks like

A handful of Pro calls outspend thousands of calls on your production model.
Your Pro-tier feature sends the same long system prompt every call, with no discount available.
Evaluation or final-pass work quietly runs on Pro because it was the best model when it was written.
One Pro conversation appears in your top spenders despite a tiny call count.

Free, no card, no account

Run the free Cost Leak Scan

It shows your most expensive conversation before you install anything.

Run the free Cost Leak Scan Start 14-day trial

Mistakes that cost the most

Mistake	Why it hurts	Do instead
Carrying a long system prompt onto Pro calls.	With no cached rate, every repeated token bills at $30 per million on every single call.	Strip the Pro prompt to what the Pro step genuinely needs, and keep the long version on cheaper tiers.
Letting Pro produce long-form output without a cap.	At $180 per million, a thousand extra output tokens costs eighteen cents on a single call.	Set a hard max-tokens ceiling on Pro calls and treat hitting it as a signal to reshape the task.
Leaving Pro in a retry path.	A retried Pro call is the most expensive single event your system can produce.	Never escalate retries to Pro automatically. Make it an explicit, logged decision.

Run this check tonight

Count your Pro calls this month and price them individually. The list is usually short and startling.
Measure how many tokens of each Pro prompt are identical across calls. That block has no discount available.
Check whether any retry or fallback path can reach a Pro model.
Compare your total Pro spend to your total production-model spend at a fraction of the call count.
Confirm each Pro call is doing work that genuinely needs it, rather than work inherited from an older design.

Twelve times the rate, and no cache to soften it

Illustrative example

A final-pass step on GPT-5.4 Pro running 2,000 times a month. Each call sends a 4,000-token prompt, of which 3,000 tokens are an identical system block, and returns 1,200 tokens.

Input: 2,000 x 4,000 = 8M tokens x $30.00 per million = $240.00

Output: 2,000 x 1,200 = 2.4M tokens x $180.00 per million = $432.00

Month total = $672.00 across just 2,000 calls, so $0.336 per call

The repeated 3,000-token block alone: 6M x $30.00 per million = $180.00, with no cached rate to reduce it

The same 2,000 calls on GPT-5.4 = (8 x $2.50) + (2.4 x $15.00) = $20 + $36 = $56

$672 for 2,000 calls against $56 on GPT-5.4, and $180 of it's a repeated prompt block that no cache can discount. On this tier prompt length is a permanent cost.

Every number, with its confidence and source

Figure	What it means	Confidence	Source
$30.00 / $180.00 per million	GPT-5.4 Pro and GPT-5.5 Pro input and output rates, identical on both models	calculated	Culpa price book rows for openai/gpt-5.4-pro and openai/gpt-5.5-pro, both $0.030 and $0.180 per 1k, effective 2026-07-02, verified against OpenAI API pricing.
12x	Pro tier cost as a multiple of GPT-5.4 on both input and output	calculated	$30.00 / $2.50 and $180.00 / $15.00, from the price-book rows.

What a generic answer can’t know

A rate card can tell you there's no cached rate. It can't tell you what share of your Pro input is a repeated block you're paying full price for on every call, because that needs comparing prompts across calls. Culpa does that on your infrastructure, where your prompts and responses stay, and counts the calls to run your plan.

Questions founders ask next

Do GPT-5.4 Pro and GPT-5.5 Pro cost the same?

Yes. Both bill $30.00 per million input and $180.00 per million output in this price book, verified 2026-07-02. Since the rate is identical, the choice between them is a capability question rather than a cost one.

Why does the Pro tier have no cached rate?

OpenAI's own pricing table shows a dash in the cached column for both Pro models, re-checked 2026-08-03, so this is the shape of the product rather than a gap in Culpa's book. The practical consequence matters more than the reason: a repeated system prompt that would cost a tenth on a cached tier costs full price here, on every call.

How do I keep Pro spend bounded?

Cap max tokens, strip the prompt to what the step needs, and make sure no retry or fallback path can reach Pro automatically. A single escalated Pro retry is the most expensive event most systems can generate.

On your infrastructure

Culpa runs on your infrastructure. Your prompts and responses never leave it. Culpa counts calls to run your plan, and it fails open, so if it ever breaks your app keeps running.

How Culpa works

Find the culprit. Not just the total.

Your dashboard shows what you spent. It stops short of who spent it. Culpa shows the conversation, the user and the feature behind it.

Your prompts stay local.

Culpa runs on your own infrastructure. What you send to a model reaches us at no point.

Every dollar has a name.

Follow any charge to the conversation, the user, the feature and the customer behind it.

See the bill before it lands.

Cost your next feature before you ship it. You get the likely bill and the worst case, at best, median, p90 and p99.

Three steps to your first answer.

Change one base URL.

Or drop in the Python or TypeScript library.

Find your most expensive conversation.

In the first session, not the first week.

Cost your next feature before you ship it.

Base URLhttp://localhost:4545/v1Your traffic keeps flowing if Culpa ever stops.

Why the bill went up

Example dashboard

Calls traced

418,209

across 3 projects

Spend this week

$378.41

+ $182 vs last week

Failed calls

312

74% retried, and you paid for all of them

+ $182 this week traced to one culprit

Spend over 14 days

user_384report_generatorconv_91fprompt_v1894,220 tokens3 retries$6.81

Most expensive users

user_384$38.42

user_119$21.07

user_562$14.90

user_204$8.30

user_871$5.10

Next week forecast

Best$180

Median$240

p90$310

p99$395

Graded against reality. Accuracy shown as results land.

Run the free Cost Leak Scan Start 14-day trial

Keep reading

Sources: OpenAI API pricing. Last reviewed 2026-08-01, rates effective 2026-07-02. Plain text version.