Anthropic has introduced Claude Haiku 5.5, the latest addition to its Claude model lineup. The company is positioning the model for high-volume and cost-sensitive workloads while also making changes to the pricing and API benefits across its model range.
Anthropic said Claude Haiku 5.5 costs around 75% less to run than Haiku 4.5 on average. It is also the company’s fastest model to date, making it suitable for workloads where response speed is important.
Claude Haiku 5.5
Claude Haiku 5.5 is the first Haiku-class model to include an adjustable effort setting. This allows users to choose whether to optimize for cost or intelligence based on their requirements.
Haiku 5.5 can handle quick and repetitive workloads such as summaries, compactions, database queries, and classification requests. It can also work as a subagent alongside Claude Opus 5.5 and Claude Sonnet 5.5 for coding work.
Performance
Anthropic has provided performance charts for Claude Haiku 5.5 across three benchmarks at each available effort setting. The comparisons show how the model performs as users adjust the effort level between cost and intelligence.
For speed-sensitive workloads, Anthropic said Haiku 5.5 is suited to applications such as live customer support and browser use.
Safety
Claude Haiku 5.5 shows major improvements across almost all of Anthropic’s alignment evaluations compared with Haiku 4.5. The company found far fewer instances of misaligned behavior and a lower willingness to cooperate with misuse. Anthropic’s system card provides more details about the evaluation process and results.
For cybersecurity, Haiku 5.5 has more restrictive safeguards than Haiku 4.5, but they are somewhat less restrictive than those applied to other recent models. The safeguards:
- Permit a wider range of defensive tasks than the safeguards used for Sonnet 5.5.
- Block penetration testing.
- Block other techniques that are more likely to be used by attackers.
Haiku 5.5 uses the same biology safeguards as Sonnet 5, Sonnet 5.5, and Opus 5. These safeguards allow research biology questions while restricting requests that Anthropic considers likely to cause harm.
Organizations working on wider-ranging biology and cybersecurity activities can apply to Anthropic’s Life Sciences Verification Program and Cyber Verification Program.
Pricing
Anthropic said Claude Haiku 5.5 costs around 75% less to run than Haiku 4.5 on average. The model is particularly suited to tasks with prompts of up to 100,000 tokens, which account for around 90% of requests to Anthropic’s previous Haiku model.
| Price per 1 million tokens | Haiku 5.5 (Up to 100k / Over 100k) | Haiku 4.5 | Sonnet 5.5 |
| Cache reads | $0.01 / $0.05 | $0.10 | $0.10 |
| Cache writes | $0.125 / $0.625 | $1.25 | $2.50 |
| Input tokens | $0.10 / $0.50 | $1.00 | $2.00 |
| Output tokens | $0.50 / $2.50 | $5.00 | $10.00 |
Further updates
Alongside the Haiku 5.5 launch, Anthropic is reducing the price of cache reads for Claude Sonnet 5.5 by 50%. The price is now $0.10 per million tokens, down from $0.20 per million tokens.
Because cache reads account for a large share of models’ token consumption, Anthropic said the change reduces the cost of Sonnet 5.5 by around 20% on most agentic tasks. The company has also provided a Terminal-Bench 4.0 chart showing Sonnet 5.5’s performance relative to cost following the price reduction.
Sonnet 5.5 and Opus 5.5 remain better choices for complex agentic coding tasks such as those measured by Terminal-Bench 4.0. Haiku 5.5, meanwhile, is intended for more narrowly scoped tasks that could have been cost-prohibitive with previous Claude versions, including compaction, summarization, and subagent work.
Anthropic is also rolling out a new monthly API credit this week for all Max and Team subscribers to use on the Claude Platform. The monthly credits are:
- Max 5x: $100
- Max 20x: $200
- Team: Up to $500, pooled across users
The credits can be used with any of Anthropic’s models to build tools, apps, and agents that call its API. Anthropic has provided further information in its Help Center.
For developers, Anthropic is updating its Claude Python and TypeScript SDKs with beta support for computer use and browser use. Further information is available in the Claude Platform documentation.
Availability
Claude Haiku 5.5 is available now on all platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure. Developers can access it on the Claude Platform using claude-haiku-5-5.