In plain English

What happened
Anthropic introduced Claude Haiku 5.5 on October 7, 2026, describing it as the fastest and most capable version of its small model line, built for high-volume work like summarization, subagents, and browser use. It follows Claude Opus 5.5 on September 22 and Claude Sonnet 5.5 on September 28, completing a refresh of all three model tiers in about two and a half weeks.
Why it matters to your business
The small model tier is usually what runs under the hood for routine, high-volume tasks: a chatbot answering common questions, a tool summarizing documents, or background steps in an automated workflow. If a business uses Claude through a product that picks the model automatically, this tier is often doing the everyday work even when nobody chose it by name.
Do this
Open the model picker in whatever tool connects to Claude and check whether it is already set to Haiku 5.5 or still running the previous version.

Anthropic's model page lists three new releases in quick succession: Claude Opus 5.5 on September 22, 2026, Claude Sonnet 5.5 on September 28, 2026, and Claude Haiku 5.5 on October 7, 2026. Each one replaces the prior version in its tier, meaning the company now has a fresh top, middle, and small model across its lineup at the same time.

What Haiku 5.5 is built for

Anthropic describes the new model as built for high-volume work like summarization, subagents, and browser use. A subagent is a smaller task handed off from a main AI assistant so it can be handled separately, rather than making the more expensive top-tier model do every step itself. Browser use refers to a model acting on a web page directly, such as following a link or filling in a form, rather than only answering questions in a chat window.

That positioning matters because small models are usually the ones running the parts of a workflow nobody notices: the automatic summary at the top of a long email thread, the background check a customer-service bot runs before it answers, the quick lookup inside a larger multi-step task. None of that requires the most powerful model available, which is the whole reason a small, fast tier exists.

The other two releases in this cycle

DateModelWhat Anthropic said changed
Sep 22, 2026Claude Opus 5.5Leads in agentic coding and knowledge work; costs 40% less to run than Opus 5 on typical workloads
Sep 28, 2026Claude Sonnet 5.5Runs 30%+ faster and costs up to 30% less than Sonnet 5 for most work
Oct 7, 2026Claude Haiku 5.5Fastest, most capable small model; built for summarization, subagents, and browser use
Anthropic's model announcements, September–October 2026

The cost figures for Opus 5.5 and Sonnet 5.5 come directly from Anthropic's own announcements. No comparable cost figure was given for Haiku 5.5 against its predecessor, Haiku 4.5.

Anthropic refreshed all three of its model tiers, Opus, Sonnet, and Haiku, within about two and a half weeks, from September 22 to October 7, 2026.

Reality check

Words like 'fastest' and 'most capable' are the vendor's own framing, not an independent measurement, and 'costs up to 30% less' applies to typical or most workloads, not necessarily the specific task a given business runs most. The only way to know whether a new model actually helps is to run the same real task through the old and new version and compare the output and the bill, not the announcement.

What to check before relying on it

  1. 1

    Find out which tier your tool actually uses

    If Claude shows up inside another product rather than directly, check that product's settings or support page to see whether it names a specific model tier, or whether it switches automatically.

  2. 2

    Look at the account's model picker

    On claude.ai or in an API account, the model selection menu will show whether Haiku 5.5 is already available or whether the account is still defaulting to an earlier version.

  3. 3

    Run one routine task through both versions

    Pick something low-stakes and repetitive, like summarizing a weekly report, and compare the new model's output against what the old one produced before switching anything that matters.

  4. 4

    Watch for a deprecation notice on the old version

    Anthropic has previously committed to preserving older models for a time rather than pulling them without notice, so there is no need to switch in a hurry.

Sources

  1. 1.Anthropic — Models