- Anthropic says Haiku 5.5 costs about 75% less to run than Haiku 4.5.
- It is the first Haiku-class model with an adjustable effort setting for cost or intelligence.
- Haiku 5.5 is available now on AWS, Google Cloud and Microsoft Azure.
Anthropic has launched Claude Haiku 5.5, which it describes as its cheapest, fastest and most capable small model so far. The model is built for high-volume and cost-sensitive work, especially quick and repetitive tasks that run in large numbers every day.
The launch was announced on October 7, 2026. According to the company, Claude Haiku 5.5 costs around 75% less to run than Claude Haiku 4.5, which makes it easier for businesses and developers to scale AI features without a steep rise in costs.
Claude Haiku 5.5 announcement
What Claude Haiku 5.5 Can Do
During evaluation, Haiku 5.5 performed better than Haiku 4.5 and GPT-6 Luna in knowledge work, computer use, multidisciplinary reasoning, agentic coding and visual reasoning. These are company-reported results, not independent benchmarks.
Also Read
Adjustable Effort Setting
Haiku 5.5 is the first Haiku-class model to come with an adjustable effort setting. Users can decide whether to optimise for cost or for intelligence, depending on the job at hand.
The company says the model is especially good value for prompts of up to 100,000 tokens. Such prompts make up around 90% of requests to earlier Haiku models, so most existing workloads fall within the lower price tier.
Safety and Alignment Changes
Anthropic reports major improvements across almost all alignment evaluations compared with Haiku 4.5. The model showed fewer instances of misaligned behaviour and a lower willingness to cooperate with misuse.
Its cybersecurity safeguards are stricter than those on Haiku 4.5 but less restrictive than those on other recent models. They allow a wider range of defensive tasks, while still blocking penetration testing and techniques more likely to be used by attackers. Biology safeguards match those on Sonnet 5, Sonnet 5.5 and Opus 5, allowing research biology questions while restricting requests that could likely cause harm.
Pricing Details
Pricing changes with prompt size. Prices below are per million tokens.
| Item | Prompts up to 100K | Prompts over 100K |
|---|---|---|
| Input tokens | $0.10 | $0.50 |
| Output tokens | $0.50 | $2.50 |
| Cache reads | $0.01 | $0.05 |
| Cache writes | $0.125 | $0.625 |
Availability and Who Should Consider It
Claude Haiku 5.5 is available now on all platforms, including Amazon Web Services, Google Cloud and Microsoft Azure. Developers can get started on the Claude Platform using the model name claude-haiku-5-5.
The model suits teams that run large volumes of simpler requests, such as classification, data extraction or customer query handling, where running cost matters more than peak reasoning power. Teams with very long prompts should compare the higher tier before scaling up.
Conclusion
Claude Haiku 5.5 targets a clear need: fast, lower-cost AI for large volumes of routine work. Its reported cost savings, adjustable effort setting and better results than Haiku 4.5 make it a practical option for developers and businesses. Before adopting it, check your typical prompt size against the two pricing tiers.
Frequently Asked Questions
What is Claude Haiku 5.5?
Claude Haiku 5.5 is Anthropic’s cheapest, fastest and most capable small model, designed for high-volume and cost-sensitive tasks.
How much cheaper is Claude Haiku 5.5 than Haiku 4.5?
Anthropic says Claude Haiku 5.5 costs around 75% less to run than Claude Haiku 4.5, on average.
What is the adjustable effort setting?
It lets users choose whether to optimise for cost or for intelligence. Haiku 5.5 is the first Haiku-class model to offer this option.
Where can developers access Claude Haiku 5.5?
It is available on Amazon Web Services, Google Cloud, Microsoft Azure and the Claude Platform, using the model name claude-haiku-5-5.
