New modelReleased 7 October 2026 · 3 min read
High-volume work just got cheap enough to try again.
Claude Haiku 5.5 is built for work that runs thousands of times a day: summaries, classification, extraction, routing. Anthropic says it costs about 75% less to run than Haiku 4.5, and it is the first Haiku with an effort setting.
At a glance
- Model
- Claude Haiku 5.5
- Price per M tokens
- $0.10 in · $0.50 out
- Context
- 1M in · 128k out
- vs Haiku 4.5
- ~75% cheaper
- Best for
- Triage, tagging, extraction
01 · What changed
Three changes on one day.
Two changes shipped beside it. Cache reads on Claude Sonnet 5.5 were cut in half, which Anthropic puts at about 20% cheaper on most agentic work. And Claude Max and Team plans now include monthly API credits for building your own tools. For a company, the practical question is which piece of repetitive work you shelved because the cost did not add up.
New model
Haiku 5.5
Cheapest, fastest small model. About 75% cheaper than Haiku 4.5, with adjustable effort.
Price cut
$0.20 → $0.10
Sonnet 5.5 cache reads per million tokens. Every other price unchanged.
Plans
Up to $500/mo
API credit on Team plans, pooled across seats. Max qualifies; Free, Pro and Enterprise do not.
02 · Where it sits
Which Claude for which job.
03 · What it means
What it means for your seat.
Most firms have a list of tasks that are dull, frequent and only half done: tagging inbound email, sorting a data room, pulling the same five fields out of every monthly report. The price is what usually stopped them.
Leadership
Re-open the business cases that died on cost. A workflow that ran the numbers on Haiku 4.5 now costs roughly a quarter as much, before you touch the design.
Operations
Haiku 5.5 fits the first pass on high-volume reading: classify, extract, route, and hand anything that needs judgment to a person or to a larger model. A covenant analysis or an IC memo is Opus 5.5’s work.
IT and engineering
Code written for Haiku 4.5 can break. Manual extended thinking, the temperature, top_p and top_k sampling parameters, and assistant message prefill can each return a 400 error on Haiku 5.5. Test before you switch a production workload.
Team plan owners
The monthly API credit is a free budget for a pilot: an internal tool, an agent, a prototype your engineer has been asking to build. It goes to one linked Console organization, and you cannot change it yourself afterwards, so pick the one you will build in.
04 · This week
What to do this week.
0 of 3 done
Sources
Facts and dates are Anthropic’s own. What it means and what to do are our view, from the firms we work with. Drafted with Claude from these sources, reviewed by Fatjon (Tony) Kalemaj and published 13 October 2026.
The short answers.
How much does Claude Haiku 5.5 cost?
From $0.10 per million input tokens and $0.50 per million output tokens on the Claude API, according to Anthropic. It runs about 75% cheaper than Claude Haiku 4.5 on average.
Is Claude Haiku 5.5 good enough for business documents?
For high-volume, well-defined work such as summaries, classification and extraction, Anthropic positions it as the right model. For work that needs sustained judgment, like reading a credit agreement or drafting an investment memo, use Claude Opus 5.5 or Sonnet 5.5. Test it on your own documents before deciding.
Do Claude Team plans include API credits?
Yes, since 7 October 2026. Team plans pool the credit across seats into one monthly balance, capped at $500. Max plans also qualify. Free, Pro and Enterprise plans do not.
Will my code for Claude Haiku 4.5 work on Haiku 5.5?
Not always. Manual extended thinking, the temperature, top_p and top_k parameters, and assistant message prefill can each return a 400 error on Haiku 5.5. Anthropic publishes a migration guide for each change.
Read next
Newsletter
Keep up with every release.
One email when something ships or retires that matters to a company: what changed, what it means, the date and what to do. Unsubscribe any time.
Want this working inside your firm?
We run Claude workshops on your own documents and stay on as your teams adopt each release.