TL;DR
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Anthropic has released Claude Haiku 5.5, a small model aimed at fast, high-volume tasks and priced substantially below Haiku 4.5. The company reports benchmark gains and improved alignment evaluations, but those results are based on its own testing; independent comparisons and broader customer results are not yet provided.
Anthropic has released Claude Haiku 5.5, a small model it says is designed for fast, high-volume tasks at lower cost than its predecessor. The launch gives developers another option for work such as summarization, classification and customer support, and includes a separate price cut for cache reads on Claude Sonnet 5.5.
Anthropic describes Haiku 5.5 as its fastest model to date and says it costs around 75% less to run on average than Haiku 4.5. Its listed token prices for prompts up to 100,000 tokens are $0.10 per million input tokens and $0.50 per million output tokens. For prompts above that length, the listed rates are $0.50 and $2.50, respectively. Anthropic says about 90% of requests to its previous Haiku model were within the lower-priced prompt range.
The model is intended for repetitive or narrowly scoped work, including summaries, database queries and classification. Anthropic also recommends it as a subagent alongside Sonnet 5.5 or Opus 5.5 on coding projects. The company says those larger models remain better suited to complex agentic coding tasks, while Haiku 5.5 may make less demanding jobs more affordable.
Haiku 5.5 is Anthropic’s first Haiku model with an adjustable effort setting, allowing users to choose between lower cost and greater reasoning effort. It is available through the Claude Platform and, according to Anthropic, through Amazon Web Services, Google Cloud and Microsoft Azure. The company also cut Sonnet 5.5 cache-read pricing by 50%, to $0.10 per million tokens, saying that change lowers costs on most agentic work by about 20%.
Lower Costs for Routine AI Work
The release is aimed at organizations that run many model requests and need quick responses without using a more expensive model for every step. Lower listed rates could make tasks such as ticket triage, summaries and data classification less costly at scale, while adjustable effort gives developers a way to manage the trade-off between response quality and expense.
Anthropic’s stated division of labor also points to a practical use: teams can reserve larger models for harder reasoning or coding and use Haiku 5.5 for simpler subtasks. That approach may reduce the cost of multi-step AI systems, though actual savings will depend on task mix, prompt size and how often requests need to be retried or escalated.
The Sonnet cache-read price reduction affects customers already using that model, rather than only those adopting Haiku. Anthropic says the cut should lower costs on most agentic tasks, but the size of the savings will vary with how much cached input a workflow uses.
As an affiliate, we earn on qualifying purchases.
Anthropic’s Model Lineup and Pricing
Haiku is the smaller, speed-focused tier in Anthropic’s model range; Sonnet and Opus are positioned for more demanding work. The company presents Haiku 5.5 as a complement to those models, not a replacement for them. Its launch materials compare Haiku 5.5 with Haiku 4.5, Sonnet 5.5 and selected competing models across knowledge work, computer use, reasoning, coding and visual reasoning.
Those benchmark results are published by Anthropic, which also points readers to a system card for details about its evaluations. Among the figures it reports, Haiku 5.5 scored 39.2% on Terminal-Bench 4.0, compared with 0.0% for Haiku 4.5 and 70.6% for Sonnet 5.5. The scores describe performance on that benchmark and should not be read as a guarantee of results in a particular company’s workflow.
Anthropic says customers in early testing reported improvements in speed and cost. Those accounts are customer reports shared by the company, not independent evaluations. The launch also includes a new monthly API credit for Claude Max and Team subscribers, which Anthropic says is intended to support development of agents and applications on its platform.
““Claude Haiku 5.5 is the cheapest, fastest, and most capable small model we’ve ever released.””
— Anthropic
As an affiliate, we earn on qualifying purchases.
Independent Results Still Pending
The available performance figures and safety findings come from Anthropic’s own evaluations; the supplied launch material does not include independent benchmark testing. The company says its alignment evaluations showed improvements over Haiku 4.5, including fewer instances of misaligned behavior and less willingness to cooperate with misuse, but the detailed methods and results are in the system card rather than fully laid out in the announcement.
It is also unclear how closely the reported cost and speed gains will carry over to different customer workloads. The stated 75% average cost reduction is a company estimate, and the savings for an individual user will depend on prompt length, output volume, caching and the selected effort setting. Anthropic has not provided a general release date for the new monthly API credit in the supplied material beyond saying it is being introduced alongside the launch.
As an affiliate, we earn on qualifying purchases.
Developers Test the New Model
Developers can begin using Haiku 5.5 through the Claude Platform and the cloud services named by Anthropic. The next useful evidence will be independent benchmark results and customer data showing how the model performs across real workloads, including whether its lower pricing translates into lower total costs once quality and retries are considered.
Anthropic directs users to its migration guide for implementation details and to the system card for evaluation and safety information. The company has not specified further model updates or a timeline for additional pricing changes in the launch material.
cost-effective AI classification software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is Claude Haiku 5.5 designed to do?
Anthropic positions it for fast, high-volume tasks such as summarization, classification, database queries and customer support. It also describes Haiku 5.5 as a possible subagent for coding workflows led by Sonnet or Opus.
How much does Haiku 5.5 cost?
For prompts up to 100,000 tokens, Anthropic lists prices of $0.10 per million input tokens and $0.50 per million output tokens. For prompts above that length, the listed rates are $0.50 for input and $2.50 for output. Other charges or platform terms may apply.
Is Haiku 5.5 available now?
Yes. Anthropic says it is available through the Claude Platform and on Amazon Web Services, Google Cloud and Microsoft Azure.
Does Haiku 5.5 replace Sonnet 5.5 for coding?
No. Anthropic says Sonnet 5.5 and Opus 5.5 remain better suited to complex agentic coding tasks. It presents Haiku 5.5 as a lower-cost option for narrower, simpler tasks within a larger workflow.
Are the reported performance improvements independently verified?
The launch figures and early customer examples in the supplied materials are reported by Anthropic. Independent test results are not included; the company points to its system card for evaluation details.
Source: hn
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
