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 agent workflows. The company says it costs about 75% less to run on average than Haiku 4.5 and reports gains on several benchmarks, while noting that Sonnet 5.5 and Opus 5.5 remain better suited to complex agentic coding.
Anthropic has released Claude Haiku 5.5, a small model the company describes as its fastest to date and says costs about 75% less to run on average than Haiku 4.5. The launch targets high-volume and speed-sensitive uses, including summaries, classification, customer support and coding subagents; Anthropic also cut cache-read prices for Sonnet 5.5.
Anthropic says Haiku 5.5 is designed for quick, repetitive, cost-sensitive work, such as summaries, context compaction, database queries and classification. It also recommends the model as a subagent alongside Sonnet 5.5 or Opus 5.5 on coding tasks, while positioning its speed for live customer support and browser use. These are the company’s intended use cases, not independent assessments of product performance.
The published benchmark table reports Haiku 5.5 scores of 72.4% on OSWorld 2.1 and 39.2% on Terminal-Bench 4.0. The OSWorld result is identified as an offline-subset score; Anthropic’s page describes OSWorld as a measure of agents’ ability to complete multi-step tasks on a real computer. On Terminal-Bench 4.0, Haiku 5.5 scores below Sonnet 5.5’s 70.6%. Anthropic says its system card provides further information about evaluation methods.
For prompts up to 100,000 tokens, Anthropic lists Haiku 5.5 pricing at $0.10 per million input tokens and $0.50 per million output tokens, compared with $1 and $5, respectively, for Haiku 4.5; Claude and OpenAI subscription value has also been examined in a study. The company says around 90% of requests to its previous Haiku model fell within that prompt range. It also says Haiku 5.5 is available on the Claude Platform and through Amazon Web Services, Google Cloud and Microsoft Azure, including via Anthropic’s AWS launch in India.
Lower Costs for High-Volume AI Tasks
The release gives developers a lower-priced option for tasks that can involve many model calls, where per-token costs and response speed affect the overall expense and user experience. Anthropic’s claim that Haiku 5.5 costs around 75% less on average than Haiku 4.5 could make some routine agent workloads more affordable, though the actual savings will depend on prompt size, output volume and how a system is built.
Anthropic also lowered Sonnet 5.5 cache-read prices by 50%, to $0.10 per million tokens. The company estimates this makes Sonnet 5.5 around 20% cheaper on most agentic work. Together, the changes adjust the costs of both a smaller model for narrower tasks and a more capable model used for complex work. The pricing could matter to teams operating agents at scale, but the company has not supplied customer-wide cost data in the announcement.
Products Worth Considering
Bose QuietComfort Wireless Bluetooth Headphones with Active Noise Cancellation, Mic, USB‑C Charging, Deep Bass, 24‑Hour Battery – White Smok
LG UltraGear 27-Inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDMI 2.1
Bose QuietComfort Wireless Noise-Cancelling Over-Ear Headphones, Black
As an affiliate, we earn on qualifying purchases.
Where Haiku Fits in Anthropic’s Model Range
Anthropic presents Haiku 5.5 as a complement to, rather than a replacement for, its larger models. The company says Sonnet 5.5 and Opus 5.5 remain better choices for complex agentic coding, while Haiku is intended for narrower tasks that may have been too costly with earlier models. Its benchmark table also shows Haiku 5.5 below Sonnet 5.5 on Terminal-Bench 4.0, a benchmark for multi-step command-line work.
Haiku 5.5 is the first model in Anthropic’s Haiku line with an adjustable effort setting, according to the company. The setting lets users choose between cost and intelligence, though the announcement does not give a single performance or price figure that applies to every setting. Anthropic also publishes a system card discussing model evaluations and safety measures.
““Claude Haiku 5.5 is the cheapest, fastest, and most capable small model we’ve ever released.””
— Anthropic
Independent Results and Cost Details
The announcement does not specify a publication date, and it does not provide independent verification of the benchmark scores, speed comparisons or average cost reduction. Anthropic refers readers to its system card for evaluation details, but results can vary with task, prompt, settings and system design. The company’s 75% average cost comparison should not be read as a guaranteed saving for every workload.
Asana’s reported latency and inference improvements come from its own early tests against the model it currently uses; the source does not identify that model or describe the test methodology in detail. The announcement also does not give a release schedule for future changes or explain the terms of the new monthly API credit for Claude Max and Team subscribers.
Developers Can Test the Model Now
Developers can access Haiku 5.5 now through the Claude Platform using the model identifier claude-haiku-5-5, or through the listed cloud providers. Anthropic directs users to its migration guide for implementation details and its system card for evaluation and safety information. Actual performance and costs will need to be tested against each application’s workload.
Anthropic says the launch also brings a new monthly API credit for Claude Max and Team subscribers to support building agents and applications on its platform. The announcement does not state the credit amount or provide further eligibility details, leaving those terms to be clarified in product information.
Key Questions
What is Claude Haiku 5.5?
Claude Haiku 5.5 is Anthropic’s small model for high-volume and speed-sensitive tasks, including summaries, classification, customer support and use as a coding subagent.
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. Larger prompts have higher listed rates. Costs depend on usage and prompt size.
How does Haiku 5.5 compare with Sonnet 5.5?
Anthropic positions Haiku 5.5 for narrower, lower-cost tasks. Its published Terminal-Bench 4.0 score is 39.2%, compared with 70.6% for Sonnet 5.5; Anthropic says its larger models are better suited to complex agentic coding.
Where is Claude Haiku 5.5 available?
Anthropic says it is available through the Claude Platform, Amazon Web Services, Google Cloud and Microsoft Azure. The platform model identifier is claude-haiku-5-5.
Source: hn
