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

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.

ByKarthik KumarCafiyn Innovations
At a glance
Model id
claude-haiku-5-5, fixed, no date suffix and no alias
Launched
7 October 2026
Context and output
1M tokens in, 128K max out
Thinking
Adaptive by default. Text omitted unless thinking.display is "summarized"
Tokenizer
About 30 percent more tokens than Haiku 4.5 for the same text
Not supported
Priority Tier. Structured outputs on Amazon Bedrock

The five requests that now fail

The migration guide is explicit that each of these returns a 400 error on Haiku 5.5. If you work from the Haiku 4.5 checklist, these are the lines to search for.

  • Manual thinking. A thinking value of {"type": "enabled", "budget_tokens": N} fails. Use {"type": "adaptive"} and set output_config.effort instead.
  • Sampling parameters. Omit temperature, top_p and top_k. If temperature is present it must be 1, if top_p is present it must be 0.99, any top_k fails, and sending both temperature and top_p fails. A top_p of 1 also fails.
  • Assistant prefill. A final assistant turn in messages fails even with thinking turned off. End with a user turn.
  • The old computer use tool. computer_20250124 fails. On the Claude API and Google Cloud use the computer_toolset_20260801 toolset, and on Amazon Bedrock use computer_20251124 with the computer-use-2025-11-24 beta header. Sending the fine-grained-tool-streaming-2025-05-14 beta header alongside a toolset entry also fails.
  • Resent thinking blocks after a change. Sending a thinking block back after changing system, tools or earlier messages fails. Keep conversations append only. On accounts created before 31 August 2026 00:00 UTC, the error comes only on requests that set thinking.block_binding.prefix_mismatch_behavior.

Thinking changes your response parsing

Adaptive thinking is on by default, so a response can begin with one or more thinking blocks even when the request never sets thinking. Code that reads the first content block as the answer will break, so select blocks by their type. The guide also says by default each thinking block comes back with an empty thinking field and only a signature, where Haiku 4.5 returned summarized thinking. Set thinking.display to "summarized" if you want the text.

Thinking tokens count toward max_tokens, so a request with a small limit can stop with stop_reason "max_tokens" after a thinking block and before any text. Raise the limit or choose a lower effort level. Thinking blocks from earlier assistant turns also stay in context and count as input tokens, where Haiku 4.5 kept only the latest turn.

A forced tool_choice of any or a named tool is accepted, but the response starts with the tool call and has no thinking block. To let the model think before a tool call, use tool_choice auto and say in the prompt when to use the tool.

Prefill has no drop in replacement

Which replacement you use depends on what the prefill was for, according to the guide.

  • Output format: use structured outputs, or tools with enum fields for classification. On Amazon Bedrock structured outputs are not available for this model, so describe the format in the prompt, use a tool without strict, and validate the output in your code.
  • Preambles: ask in the system prompt for a direct answer.
  • Continuations: move them into the user message, for example by saying the previous response was interrupted and quoting where it ended.
  • Context reminders: put them in the user turn.

What fails quietly, or only on some platforms

The tokenizer is the one that reaches your bill and your limits: about 30 percent more tokens for the same text, and a 2,000 by 1,500 pixel image costs about 2.5 times as many visual tokens because Haiku 5.5 uses the high resolution image tier. A max_tokens tuned for Haiku 4.5 can truncate equivalent output. See our pricing story for how that interacts with the 100,000 token price line.

Thinking blocks from Haiku 5.5 work only in the account that produced them or an account linked to it. When another account sends one, the API drops the block and the request succeeds without that reasoning, so a service that replays stored conversations through a different account loses context without an error.

Haiku 5.5 runs safety classifiers that can decline a request, and it has no server side fallback, so handle stop_reason "refusal". If your organisation has a Priority Tier commitment on Haiku 4.5, plan that capacity separately because Priority Tier is not supported on Haiku 5.5.

Managed Agents users are the exception: the guide says no changes beyond updating the model name are required.

What to do

  1. 1Replace the model id with claude-haiku-5-5 for your platform. Amazon Bedrock uses anthropic.claude-haiku-5-5.
  2. 2Search for budget_tokens, temperature, top_p, top_k and any request that ends with an assistant message, and remove or rewrite each.
  3. 3Select response content blocks by type, and raise max_tokens to leave room for thinking.
  4. 4Recount your prompts with model set to claude-haiku-5-5, then recompute cost and limits.
  5. 5Handle the refusal stop reason, and keep conversations append only if you send thinking blocks back.
  6. 6Move computer use to computer_toolset_20260801, or computer_20251124 on Bedrock.

Common questions

What is the Claude Haiku 5.5 model id?

claude-haiku-5-5 on the Claude API, Claude Platform on AWS, Google Cloud and Microsoft Foundry, and anthropic.claude-haiku-5-5 on Amazon Bedrock. It is a fixed id with no date suffix and no alias.

Can I still set temperature on Haiku 5.5?

Only to its default. If temperature is present it must be 1, and a request that sets both temperature and top_p returns a 400 error. Anthropic recommends omitting these and using prompting instead.

Does prefill work with thinking off?

No. Haiku 5.5 rejects an assistant prefill with a 400 error even when thinking is turned off.

Do I need to change anything for Managed Agents?

The migration guide says if you use Claude Managed Agents, no changes beyond updating the model name are required.

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 100,000 token price line →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.
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.
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.
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