Claude Haiku 5.5

What Claude Haiku 5.5 is

Claude Haiku 5.5 is a cloud model positioned for high-volume tasks such as classification, extraction and routing. Classification assigns an item to a category; extraction pulls specified information from supplied material. These are narrower tasks than asking an agent to manage an entire open-ended project.

A short, explicit output format makes those tasks easier to check. For example, extracting accepted commitments from a meeting is different from writing a general summary or inventing a helpful project plan. The model must distinguish what someone actually promised from a tentative suggestion.

How it fits into a workflow

The meeting workshop supplies a fictional transcript containing two commitments, a tentative offer and later corrections. The proposed workflow asks for a preview with evidence timestamps before producing a CSV. A person checks the two accepted tasks and keeps unresolved questions separate. This is a documented practice recipe, not a measured model run.

Examples described as illustrative or proposed trials are editorial suggestions. Documented real builds, when available, are linked separately below.

Requirements, cost and limitations

Check the relevant Claude product or API access and current pricing. Prompt-length tiers and a newer tokenizer affect billed usage, so a lower price per token is not necessarily a lower price per completed task. The documented context and output limits do not replace task-level evaluation.

When to choose it and what to compare

Evaluate Haiku when the task is bounded and repeated, with outputs you can validate. A more complex or interdependent task may justify testing Sonnet or another model. A simple rule or spreadsheet may be preferable when the input is already structured and no language interpretation is needed.

Access, running cost and practical requirements

Efficiency-focused hosted option. How we use this label.

Claude API and documented partners. As of October 29, Haiku 5.5 replaces Haiku 4.5 in Free/Pro/Max chat, Cowork and Design model selectors, including desktop/mobile.

Developer API prices

Billing route

Claude API base

Units

USD per one million tokens

Input

$0.1

Cached input

$0.01

Output

$0.5

These rates require prompts up to 100,000 tokens. Above 100,000, input $0.50, cached input $0.05, output $2.50. The prompt-length tier affects the applicable rate, not simply a flat 'from' price.

API prices are not a chat subscription price. A token is a unit of processed text; images and other inputs have their own token accounting. Repeated history, reasoning output and tools can change the final bill. A subscription may offer separately claimed credits, subject to its terms.

What the configuration changes

Adaptive thinking with default API effort medium.

Narrow repetitive tasks can be cheaper to evaluate than delegating an entire open-ended project.

New tokenizer counts roughly 30% more tokens than Haiku 4.5 for the same text; workload-dependent.

Product context differs: paid chat/Code list 1M; Cowork lists 500K. Do not apply API ceiling to every interface.

Benchmark evidence and what it means

These are selected source-reported observations, not a league table. Task, version, tool access, reasoning budget and test harness all affect a result. A historical score does not establish current leadership or accuracy on your own work.

Sources and evidence

Product facts are based on the following published sources, checked 2026-10-27. Workflow suggestions and comparisons are editorial interpretation; PrisAIm has not independently benchmarked this product.