Near-instant responses
Time to first token measured in fractions of a second, so chat, autocomplete, and voice flows feel live.
Now available · Claude Haiku 5.5
The fastest Claude, ready for every request.
Haiku 5.5 pairs near-instant responses with strong reasoning, vision, and tool use. Build assistants, pipelines, and products that feel immediate, at a price that scales.
- < 0.5stypical time to first token
- Visiontext and image input
- Toolsstructured function calls
Features
Everything you need to ship faster.
Haiku 5.5 is tuned for the workloads where speed and cost decide whether a feature gets built at all.
Built for high volume
Run thousands of classification, extraction, and routing calls per minute without blowing your budget.
Vision input
Read screenshots, scanned documents, charts, and photos alongside text in a single request.
Reliable tool use
Call functions and APIs with structured JSON arguments, ideal for agents that act on your systems.
Strong at code
Write, explain, and refactor code across common languages, with fast iteration on small diffs.
Safety by design
Trained with Anthropic’s safety practices, with clear controls for system prompts and content policies.
Where Haiku fits
Pick the right model for every step.
Route simple, high-frequency work to Haiku 5.5 and save larger models for the hard problems. Your users get fast answers, and your bill stays predictable.
- Customer support triage and reply drafting
- Document extraction into structured records
- Real-time moderation and content routing
- In-product assistants and search rewriting
- Sub-agents inside larger multi-model pipelines
Relative latency
Lower is faster. Illustrative values for comparison.
Pricing
Simple plans, pay-as-you-go API.
Start free. Upgrade when you need more headroom or want to build with the API.
Free
$0forever
Explore Haiku 5.5 in chat with generous daily limits.
- Chat access on web and mobile
- Daily usage limits
- File and image uploads
- Community support
Pro
$20per month
Higher limits and priority access for individuals who work with AI daily.
- Everything in Free
- 5× higher usage limits
- Priority access during peak hours
- Early access to new features
- Email support
API
Pay as you goper million tokens
Build Haiku 5.5 into your product with usage-based billing.
- Input: $1.00 / MTok
- Output: $5.00 / MTok
- Prompt caching and batch discounts
- Usage dashboards and rate-limit tiers
- Developer documentation and SDKs
Indicative pricing for this landing page. Confirm current rates on Anthropic’s pricing page before publishing.
FAQ
Questions, answered.
How does Haiku 5.5 compare to larger Claude models?
Haiku 5.5 is the fastest and most cost-efficient model in the family. Choose it for high-throughput, latency-sensitive work; reach for a larger model when a task needs deeper multi-step reasoning.
Can I try it before paying?
Yes. The Free plan lets you chat with Haiku 5.5 with daily limits, and API customers can start with pay-as-you-go billing and no monthly commitment.
Does it support images and tools?
Yes. Haiku 5.5 accepts image input and supports tool use, so you can connect it to your own functions, databases, and services.
Are the prices final?
The prices shown here are indicative for this page. Confirm current rates on Anthropic’s pricing page before publishing.
Start building with Haiku 5.5 today.
Free to try. Scale to millions of requests when you are ready.