Guides / gpt-5.6 terra pricing
GPT-5.6 Terra pricing against the model it replaces
GPT-5.6 Terra pricing is $2.00 per million input tokens and $12.00 per million output tokens, which undercuts GPT-5.4 on both sides at $2.50 and $15.00. Output bills six times input, the usual OpenAI shape. Culpa, a local-first LLM cost, margin, and forecast ledger, records the model on every call, so a migration lands as a measured delta.
Why this happens
Terra sits where GPT-5.4 has been the production default, and it prices 20% below it on both input and output. That makes the migration arithmetic unusually clean, because the six-to-one ratio is unchanged and only the absolute rate moves. The catch is behavioural rather than financial. A model swap changes outputs, so the saving is only real if quality holds on your own data, and most teams never measure that either side of the change.
What this usually looks like
- Your production default is still GPT-5.4 and nobody has priced the Terra alternative.
- A migration got discussed on capability grounds with no cost figure attached either way.
- Spend per feature hasn't been recomputed since the newer tier appeared.
- Nobody can say what a 20% rate cut is worth on your actual token volume.
Free, no card, no account
Run the free Cost Leak Scan
It shows your most expensive conversation before you install anything.
Mistakes that cost the most
| Mistake | Why it hurts | Do instead |
|---|---|---|
| Migrating on the headline rate without re-testing output quality. | A cheaper model that fails more often carries a higher effective cost, so the saving can reverse. | Run both on a hundred real inputs for a week and compare effective cost, not list cost. |
| Assuming the 20% cut applies evenly across your bill. | It applies per token, so a feature dominated by retries or untagged background work barely moves. | Apply the new rates to your own input and output split before promising anyone a number. |
| Leaving the old model pinned in paths nobody reviews. | Background jobs and eval suites keep paying the older rate long after the interactive path moved. | List every call site by model, then migrate deliberately rather than one surface at a time. |
Run this check tonight
- Pull your GPT-5.4 input and output token totals for the last 30 days, separately.
- Re-price them at $2.00 and $12.00 per million. That difference is what the migration is worth.
- Check whether any eval or background path is pinned to the older model.
- Compare failure rates on both models over the same inputs before committing.
- Confirm the cached rate is in use. Terra carries $0.20 per million, a 90% discount.
What a Terra migration is worth on real volume
Illustrative example
A feature sending 20 million input and 4 million output tokens a month, currently on GPT-5.4 at $2.50 and $15.00 per million.
The migration alone is 20%. Caching the repeated input on top takes it to 40%, and that second half is available whether you migrate or not.
Every number, with its confidence and source
| Figure | What it means | Confidence | Source |
|---|---|---|---|
| $2.00 / $12.00 per million | GPT-5.6 Terra input and output rates | calculated | Culpa price book row for openai/gpt-5.6-terra, $0.002 and $0.012 per 1k, read off the OpenAI pricing page on 2026-08-01. Pending seed into the ledger migrations. |
| 20% | Terra discount against GPT-5.4 on both input and output | calculated | ($2.50 - $2.00) / $2.50 and ($15.00 - $12.00) / $15.00, from the two price-book rows. |
What a generic answer can’t know
A rate card can tell you Terra is 20% below GPT-5.4. It can't tell you whether your outputs got worse, or which call site is still pinned to the old model months later. That needs per-call model attribution joined to your own quality signal. Culpa records it on your infrastructure, keeps your prompts and responses there, and counts the calls to run your plan.
Questions founders ask next
How much does GPT-5.6 Terra cost?
$2.00 per million input tokens and $12.00 per million output, with cached input at $0.20. That's 20% below GPT-5.4 on both sides, and the six-to-one output ratio is unchanged.
Is GPT-5.6 Terra cheaper than GPT-5.4?
Yes, by 20% on both input and output. Whether your bill falls by 20% depends on your token split and how much of your spend is retries or untagged background work, which a rate change doesn't touch.
Should I migrate from GPT-5.4 to Terra?
Price it first, then test quality on your own inputs. A cheaper model that fails more often costs more in effective terms, so the list-rate saving is the start of the decision rather than the end of it.
Does the $2.00 rate hold on long prompts?
No. OpenAI publishes a second, long-context tier where input doubles and output rises by half, and the rate above is the short-context one. That tier covers every full-size OpenAI model rather than this family alone. Culpa's price book seeds the short-context tier, so a long-context call prices as a floor rather than an overcount. Re-checked 2026-08-03.
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.
Why the bill went up
Example dashboardCalls 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
Most expensive users
Next week forecast
Graded against reality. Accuracy shown as results land.
Keep reading
Sources: OpenAI API pricing. Last reviewed 2026-08-01, rates effective 2026-08-01. Plain text version.