Guides / claude sonnet 5 pricing
Claude Sonnet 5 pricing, and the rate change on 2026-09-01
Claude Sonnet 5 pricing is $2.00 per million input tokens and $10.00 per million output tokens on an introductory rate. From 2026-09-01 it rises to $3.00 and $15.00, a 50% increase on both sides. Culpa, a local-first LLM cost, margin, and forecast ledger, holds dated rates, so a forecast built today prices September correctly.
Why this happens
Sonnet 5 is the only model in this price book carrying two rate rows with different effective dates. That makes it the clearest example of why a price book needs versions rather than a single current number. Any forecast built on today's $2.00 rate under-reads September by a third, and any attempt to re-price August's history at the September rate rewrites the past. Both failures are silent.
What this usually looks like
- Your September forecast was built on the August rate and nobody flagged the change.
- Historical cost reports shift when re-run, because the pricing lookup uses the current rate rather than the dated one.
- The budget for a Sonnet 5 feature was signed off against introductory pricing with no end date noted.
- Nobody can say which rate a given month's spend was calculated at.
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 |
|---|---|---|
| Forecasting September at the introductory rate. | The forecast under-reads by a third, and the miss lands in the first month anyone checks. | Price forward at $3.00 and $15.00 from 2026-09-01, and say in the forecast which rate each month used. |
| Re-pricing historical calls at the current rate. | It rewrites months that were genuinely cheaper, so trend analysis becomes meaningless. | Price every call at the rate in effect on its own date, and keep the effective date on the record. |
| Assuming an introductory rate is permanent because nothing announced its end. | The end date exists in the price book already, so the increase is scheduled rather than a surprise. | Check the effective dates on every rate you depend on, and diary the ones that change. |
Run this check tonight
- Confirm which rate your cost system uses for August spend and which it will use for September.
- Re-run a month-old report and check the total still matches. A shift means dated pricing isn't in place.
- Price your current Sonnet 5 volume at $3.00 and $15.00. That's your September bill at flat usage.
- Check whether any budget or customer quote was built on the introductory rate.
- Look at whether Sonnet 4.6 is now the same price as Sonnet 5 will be, and whether that changes your choice.
The same usage, priced either side of 2026-09-01
Illustrative example
A production feature sending 20 million input and 4 million output tokens a month, unchanged across the rate boundary. Both rates come from the same model's two price-book rows.
A 50% increase arrives on 2026-09-01 with no usage change and no action on your side. It's in the price book today, so there's no excuse for it to surprise a forecast.
Every number, with its confidence and source
| Figure | What it means | Confidence | Source |
|---|---|---|---|
| $2.00 / $10.00 per million | Claude Sonnet 5 introductory input and output rates, effective 2026-07-02 | calculated | Culpa price book row for anthropic/claude-sonnet-5, $0.002 and $0.010 per 1k, effective 2026-07-02, verified against Anthropic pricing. |
| $3.00 / $15.00 per million | Claude Sonnet 5 standard rates, effective 2026-09-01 | calculated | Second price book row for anthropic/claude-sonnet-5, $0.003 and $0.015 per 1k, effective 2026-09-01. |
| 50% | rate increase on both input and output at the 2026-09-01 boundary | calculated | ($3.00 - $2.00) / $2.00, and ($15.00 - $10.00) / $10.00, from the two dated rows above. |
What a generic answer can’t know
A pricing page shows today's rate. It can't re-price your own history at the rate that was live when each call ran, which is what makes a cost trend trustworthy. That needs a versioned price book joined to dated call records. Culpa keeps both on your infrastructure, where your prompts and responses stay, and counts the calls to run your plan.
Questions founders ask next
When does Claude Sonnet 5 introductory pricing end?
The price book carries the introductory rate from 2026-07-02 and a standard rate effective 2026-09-01, so the introductory period runs through 2026-08-31. Verify against Anthropic's pricing page before committing a budget to it.
How much does the Sonnet 5 rate change cost me?
50% on both input and output, from $2.00 and $10.00 to $3.00 and $15.00 per million. At 20 million input and 4 million output tokens a month that's $40 more, with no usage change.
Should I re-price old months at the new rate?
No. Price every call at the rate in effect on its own date. Re-pricing history at a current rate makes cheaper months look expensive and destroys any trend you were trying to read.
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: Anthropic pricing. Last reviewed 2026-08-01, rates effective 2026-07-02. Plain text version.