Cafiyn Pulse
← Startup Signals
6 min readUpdated 2026-10-15Issue 6

Claude Haiku 5.5 costs a tenth of Haiku 4.5 per token. One token past 100,000 it costs half.

The rate card is a tenth of Haiku 4.5. The footnote is a threshold that moves the price fivefold and that a tokenizer change pushes you toward.

ByKarthik KumarCafiyn Innovations
At a glance
Up to 100,000 prompt tokens
$0.10 input, $0.50 output per million tokens
Over 100,000 prompt tokens
$0.50 input, $2.50 output per million tokens
Cache read
$0.01 up to the line, $0.05 over it
Haiku 4.5, for comparison
$1 input, $5 output, $0.10 cache read
Batch
$0.05 and $0.25 up to the line, $0.25 and $1.25 over it
Tokenizer
About 30 percent more tokens than Haiku 4.5 for the same text
Haiku 4.5 status
Active. Tentative retirement "Not sooner than October 15, 2026"

The rate card, and the line

Anthropic launched Claude Haiku 5.5 on 7 October. The pricing page lists two sets of prices for it. For prompts up to 100,000 tokens, input is $0.10 and output is $0.50 per million tokens, with 5 minute cache writes at $0.125, one hour cache writes at $0.20 and cache reads at $0.01. For prompts over 100,000 tokens, input is $0.50, output is $2.50, and the cache figures are $0.625, $1 and $0.05. Haiku 4.5 is $1 and $5.

So below the line the new model is a tenth of the old one on both sides. Above it, it is half. Claude 4.6 and later models "except Claude Haiku 5.5" include the full 1M token window at standard pricing, one rate across the whole window, so this is a different shape of pricing for Anthropic.

What counts toward the line

The pricing page defines it: "A request's prompt length counts all of its input tokens, including cache reads and cache writes." Each request is priced on its own, and a request over the threshold pays the higher prices "even when part of its prompt is a cache hit", while earlier requests keep the prices they were billed at.

Two consequences follow. Caching does not get a long prompt under the line, because cached tokens still count toward its length. And an agent loop crosses the line mid session: the early turns bill at the low rate and the later turns, which carry the accumulated context, bill at the high one.

Three requests, worked through

These use the stated rates and the stated 30 percent token increase, which Anthropic describes as approximate and dependent on content. Treat them as shapes, not forecasts.

  • A 50,000 token prompt and 2,000 tokens of output on Haiku 4.5 costs $0.05 plus $0.01, so $0.06. The same text on Haiku 5.5 is about 65,000 input tokens and 2,600 output tokens: $0.0065 plus $0.0013, so $0.0078. That is 13 percent of the old cost.
  • A 120,000 token prompt with 2,000 tokens of output on Haiku 4.5 costs $0.12 plus $0.01, so $0.13. On Haiku 5.5 it is about 156,000 input tokens at the higher rate, $0.078, plus 2,600 output tokens at $2.50, $0.0065, so about $0.0845. That is 65 percent of the old cost.
  • The edge: a prompt of 76,000 tokens by Haiku 4.5 counting becomes about 98,800 on Haiku 5.5 and costs roughly $0.0099 on the input side. A prompt of 78,000 becomes about 101,400, crosses the line, and costs roughly $0.0507. About 2.6 percent more text costs about five times as much.

The tokenizer pushes you toward the line

The migration guide says Haiku 5.5 uses the same tokenizer as Claude 4.7 and later, and that the same text produces approximately 30 percent more tokens than on Haiku 4.5. If that holds for your content, a prompt of about 77,000 tokens by Haiku 4.5 counting lands at the 100,000 token line on Haiku 5.5. Count with the model set to claude-haiku-5-5, as the guide says, rather than reusing old counts.

Thinking adds to it. The guide says thinking blocks from earlier turns stay in context and count as input, where Haiku 4.5 kept only the latest turn. A multi turn conversation therefore grows faster toward the line than the tokenizer change alone suggests.

How it compares, and what not to rush

OpenAI lists GPT-6 Luna at $0.10 input and $0.50 output for short context prompts, the same headline rates, with the short context limit at 272K input tokens according to its changelog. Above that the pricing page charges $0.20 and $0.75, so input doubles and output rises by half, where Haiku 5.5 steps up fivefold on both. Batch pricing for the two matches below their lines, at $0.05 and $0.25.

There is also no deadline forcing the move. The deprecations page lists Haiku 4.5 as active with a tentative retirement of "Not sooner than October 15, 2026", no deprecation has been announced, and Anthropic says it provides at least 60 days of notice before retiring a publicly released model. If you pin US only inference with inference_geo, the 1.1x multiplier applies to the higher prices too.

What to do

  1. 1Count a sample of real prompts with model set to claude-haiku-5-5, and look at the 95th percentile, not the mean.
  2. 2Find the share of requests that would land over 100,000 tokens, and cost that share at $0.50 and $2.50.
  3. 3For agent loops, measure prompt length by turn, because thinking blocks from earlier turns stay in context.
  4. 4Decide what happens to long prompts: trim them, summarise history, or route them to a different model.
  5. 5Keep Haiku 4.5 in place until the numbers justify the move. Nothing has been deprecated.

Common questions

How much does Claude Haiku 5.5 cost?

$0.10 input and $0.50 output per million tokens for prompts up to 100,000 tokens, and $0.50 and $2.50 for prompts over 100,000 tokens, per the Anthropic pricing page.

Does prompt caching keep me under the 100,000 token line?

No. The pricing page says prompt length counts all input tokens including cache reads and cache writes, and that a request over the threshold pays the higher prices even when part of the prompt is a cache hit.

Is the whole request billed at the higher rate once it crosses?

The pricing page says a request is priced on its own by its prompt length, and that a request over the threshold pays the higher prices. It does not describe a blended rate.

Is Haiku 4.5 being retired?

Not yet. The deprecations page lists it as active with a tentative retirement of not sooner than 15 October 2026, and Anthropic provides at least 60 days of notice before retiring a publicly released model.

Sources

Read against the primary documentation rather than secondary coverage. Where a figure comes from a provider's own docs, it is quoted as published on 2026-10-15.

Related

The Haiku 5.5 migration, request by request →The Sonnet 5.5 migration →Claude Haiku 4.5 pricing →

More from Signals

OpenAI removes o1, o3-mini, o4-mini and GPT-4 Turbo on 23 October. The replacements it names run from 5.5 times the price to 97 percent less.
Eleven models and six fine tuned models go dark in four days. Here is OpenAI's own mapping, what each landing costs on a typical month of traffic, and the places where the named replacement is not the cheapest or cannot be used at all.
OpenCode says a breach hit 1.4 percent of Console users who linked an AI provider. Rotate those keys before the root cause is published.
OpenCode has confirmed the breach and has not said how it happened or what was taken. Here is what the vendor said, who is exposed, the order to work in, and how to limit the damage next time.
GitHub Actions workflow lockfiles are in public preview. Here is how to turn them on, check them in CI, and what they do not protect.
A tag can move after you reviewed it. A lockfile makes the runner use the commit you reviewed. Here is the setup, the CI check, the policy, and what a lockfile cannot tell you.
Next.js 16.4.1 and 15.5.28 fix two critical remote code execution flaws. For one of them there is no workaround.
Two of the three fixes were postponed from September. Here is what each flaw needs in order to be reachable, the one workaround that exists, and what the announcement leaves out.
A poisoned npm release that punishes you for revoking the stolen token. The order of your response is the whole defence.
Most incident guidance starts with revoking the leaked token. Here that is the trigger. What StepSecurity found, the order it recommends, and why a valid provenance attestation did not help.
Gemini 4 Argon launched at $2 and $10 and goes to $4 and $20 on 1 December. Price your product on the second number.
Every tier on the pricing page doubles on the same date. What changes, what the number includes, and what to do if you are building on the launch rate.
Claude Haiku 5.5 is not a model id swap. Five things that worked on Haiku 4.5 now return a 400.
The model id is the easy part. The requests that now fail, the response parsing that changes, and the details that fail quietly.
Next.js shipped 16.3.7 and 16.3.8 a day apart. Only one of them is the security release.
Two patch releases a day apart, one of them a decoy if you are not reading carefully. And the 15.5 line does not get the high severity fix at all.
The EU AI output marking rule is not a December deadline. For most teams it started in August.
Almost every summary of this gives one date. There are two, and which one applies to you depends on when you shipped, not on how big you are.
Sonnet 5.5 cut the rate by a third and the token count went up by about 30 percent.
A third off the per-token rate is not a third off the bill when the same text now costs more tokens. Here is the arithmetic, and the five ways your code can break.
Your AI gateway holds every provider key you have. A 9.9 just landed in LiteLLM.
An AI gateway concentrates every upstream key into one process. That is the point of it, and it is why a privilege escalation there is worse than it looks.
The table your coding agent created does not have row level security on it.
Dashboard-created tables get row level security by default now. API-created tables do not, and API is how your coding agent works.
A 9.5 in next/og. Whether it reaches you depends on one line in your OG route.
The advisory rates it 9.5, but it only fires if your OG image route puts attacker-controlled text into SVG. Here is how to find out which side of that line you are on.
OpenAI and Anthropic both cut prices on 22 September. One of them costs you a refactor.
Two frontier price cuts in one day is a good week for anyone paying a token bill. The migration cost is not evenly distributed, and one of the two prices comes with an expiry hedge.
A package install reads your home directory. This week two of them did it on purpose.
The implant did nothing clever. It read the files your shell reads, in the place your credentials live, and sent them somewhere else. That is the whole attack, and it works.
GitHub is switching off SHA-1 SSH keys. Find out now whether yours is one of them.
The keys at risk are the old ones, on the machines nobody logs into, in the pipelines that have run untouched for three years. Those are also the ones that will fail loudest.
Every third-party script tag is a standing grant to run code on your visitors.
A marketing form, a chat widget, an analytics snippet. Each one is permission for someone else's server to execute whatever it sends, on every page it sits on, forever.
1.8 million Android apps were scanned for hardcoded keys. Here is how to check yours.
Finding secrets in shipped apps is now a cheap, automated pipeline. Anything compiled into an APK, an IPA or a JavaScript bundle should be treated as already published.
OpenAI's Agents API will hold your agent state. What breaks if you ever need to leave.
Two major providers moved into the agent state layer in the same week. A managed harness is a real shortcut, as long as you keep a copy of what it holds.
Meta's Muse agent can now check out at your store. What merchants should change.
Agent purchases are arriving through wallets and single-use cards, not through merchants adopting a new protocol. Many merchants will receive them without choosing to.
What Brevo's SSO incident teaches about multi-tenant SAML and hijacked sending accounts.
One incident, two lessons: why email authentication did not help, and the tenant-scoping mistake anyone shipping SSO can make.
Cloudflare now sorts bots into Search, Training and Agent. What it means for your site and your agent.
The same change lands on two audiences. Site owners can now accidentally block search; agent builders will find more doors closed.
OpenAI has announced ten dated deadlines between 23 October 2026 and 1 April 2027. Here is the calendar.
Ten dated deadlines in six months, every one announced in public. The failure mode is never the notice. Updated 19 October with the entries added since September.
Nvidia bought Hugging Face. Your open-weight fallback is now a vendor relationship.
The commitments say nothing breaks. The structural change is that "we can always self-host" is now a dependency on one vendor rather than on a neutral commons.
The Videos API goes away on 24 September and OpenAI names nothing to move to.
Every other deprecation on OpenAI's page offers a migration target. This one does not, which makes it a different kind of problem.
Whisper shuts down on 26 February. The risk is not the deadline.
Five months is a generous window. Whisper is also typically wired into a background job that fails into a queue rather than into a user's face.
The Assistants API is gone. Your app may not have told you yet.
It shut down on 26 August 2026. The replacement is not a drop-in, and the most common migration mistake changes your model's behaviour without throwing an error.
One Copilot code review costs 13 requests. Here is the rest of the maths.
Copilot now bills on usage rather than seats. The headline change is simple. The multipliers underneath it are where teams get surprised.
Row-level security is on by default now. That is not the same as being protected.
Lovable fixed the default. The gap that actually leaks data was never the default, it was the policy, and no platform can write that for you.
Your model comparison is measuring the wrong number.
Every price list is per token. Your bill is per finished piece of work. Those two numbers can rank models in opposite orders.
An agent with production log access is a new identity. Treat it like one.
Always-on agents that read production logs are arriving fast. The access-control question they raise is old and well understood, which is the good news.

Open the tool.

Ten production categories, three minutes, and a ranked list of what to fix first.

Check your own stack