Is it good for coding?

It works well as a subagent alongside Opus 5.5 or Sonnet 5.5 on coding work (scoring 46.4% on FrontierCode 1.1 Main), but Anthropic states Sonnet 5.5 and Opus 5.5 remain better for complex agentic coding tasks like Terminal-Bench 4.0.

No. Claude Haiku 5.5 is a proprietary Anthropic model accessed via the Claude Platform and partner cloud providers.

How does it compare to Sonnet 5.5 and Opus 5.5?

Haiku 5.5 is much cheaper and faster but less capable on complex tasks. Anthropic positions it for narrowly scoped, high-volume work that would have been cost-prohibitive with previous Claude versions, while Sonnet 5.5 and Opus 5.5 handle demanding agentic coding.

Do Claude subscription plans include Haiku 5.5?

Yes — the plan comparison shows Haiku model access on Free, Pro, and Max tiers, and Max and Team subscribers receive monthly API credits ($100–$500 depending on tier) usable with any model, including Haiku 5.5, on the Claude Platform.

Claude Haiku 5.5 — Anthropic's fastest, most capable small AI model

Claude Haiku 5.5 is Anthropic's smallest and fastest model, positioned for high-volume, cost-sensitive workloads such as summarization, compaction, classification, database queries, subagent tasks, live customer support, and browser use. It is the first Haiku-class model with an adjustable effort setting, letting developers trade cost against intelligence, and is priced roughly 75% lower on average than Haiku 4.5 (90% cheaper for prompts up to 100k tokens, 50% cheaper above that).

Key Features

Low per-token pricing

Input tokens cost $0.10 per million (prompts up to 100k tokens) or $0.50 above that; output tokens cost $0.50 / $2.50 respectively, with cache reads at $0.01 / $0.05 per million tokens.

Adjustable effort setting

First Haiku-class model with Low, Medium, High, Xhigh, and Max effort levels, letting users optimize for cost or intelligence per task.

Strong benchmark gains over Haiku 4.5

Scores 72.4% on OSWorld 2.1 (computer use, offline subset) vs. 15.7% for Haiku 4.5, 45.9% on Humanity's Last Exam without tools vs. 10.2%, and 39.2% on Terminal-Bench 4.0 agentic coding.

Subagent and coding support

Designed to pair with Opus 5.5 and Sonnet 5.5 as a subagent on coding work; Cognition reports a top-tier FrontierCode score of 66.2 in Devin Fusion with Haiku 5.5 as the sidekick.

Computer use and browser use in beta

Updated Claude Python and TypeScript SDKs add beta support for computer use and browser use, tasks Haiku 5.5 is suited to given its speed, capability, and price.

Multi-cloud availability

Available now on Amazon Web Services, Google Cloud, and Microsoft Azure, and via the Claude Platform under the model ID claude-haiku-5-5, with a migration guide provided.

Improved alignment and safeguards

Shows fewer misaligned behaviors and lower willingness to cooperate with misuse than Haiku 4.5; cybersecurity safeguards permit defensive tasks but still block penetration testing, and biology safeguards match those of Sonnet 5/5.5 and Opus 5.

1M-token context window

The pricing page lists a context window of up to 1 million tokens, varying by model, with Haiku access included on all consumer plans.

How It Works

Model ID on the Claude Platform

Developers access the model via the claude-haiku-5-5 model ID on the Claude Platform, with a migration guide for teams moving from Haiku 4.5.

Effort-based cost tuning

For each task, developers choose an effort setting (Low through Max) that trades accuracy on benchmarks like OSWorld, GDPval-AA, and Humanity's Last Exam against per-attempt cost.

Beta SDK support for computer/browser use

Python and TypeScript SDKs add beta computer use and browser use support, enabling speed-sensitive agent workloads such as browser automation and live support.

Pros & Cons

Pros

Dramatically cheaper than its predecessor: ~75% cheaper on average, and 90% cheaper for prompts under 100k tokens
Fastest Claude model to date at standard speed, with customer-reported latency reductions (e.g., 30% lower task-completion latency at Asana)
Substantial benchmark improvements over Haiku 4.5 across computer use, reasoning, and coding
Adjustable effort setting gives fine-grained control over the cost/intelligence tradeoff
Available on all three major clouds plus the Claude Platform
Improved alignment results and explicit, documented safeguard policies

Who It's For

Best for

Developers running high-volume, cost-sensitive workloads like summarization, classification, and compaction
Teams using Opus 5.5 or Sonnet 5.5 for coding who want a fast, cheap subagent
Latency-sensitive applications such as live customer support and browser use
Enterprises doing analytical work at scale (financial summaries, portfolio screening, CRM audits)
Max and Team subscribers who want to experiment with API-built agents using included monthly credits

Not ideal for

Complex, multi-step agentic coding tasks better served by Sonnet 5.5 or Opus 5.5

Use Cases

High-volume summarization and compaction

Classification and quick lookups

Database queries and document Q&A (e.g., AlphaSense's 8M weekly 'Ask in Document' calls)

Subagents pairing with Opus 5.5 or Sonnet 5.5 on coding work

Live customer support and other latency-sensitive tasks

Browser use and computer use agents

Analytical work at scale such as cost reports, financial summaries, and CRM record audits

Pricing

API pricing is per million tokens: input $0.10 (prompts up to 100k tokens) / $0.50 (over 100k), output $0.50 / $2.50, cache reads $0.01 / $0.05, cache writes $0.125 / $0.625. This is roughly 75% cheaper on average than Haiku 4.5 ($1.00 input / $5.00 output). Separately, Claude consumer plans include: Free at $0, Pro at $17/month billed annually ($200 up front; $20 billed monthly), and Max from $10

Free

Chat on web, desktop, and mobile; web search, file creation, and code execution; memory across conversations; access to Haiku, Sonnet, and Claude Design/Slides/Docs; up to 5 Projects.

Pro

$17/month (annual, $200 billed up front; $20 monthly)

Everything in Free plus more usage, task handoff and scheduling, Claude Dashboards (beta), Claude Code, Claude Science, Projects, more models, and Claude in Chrome and Microsoft 365.

Max

From $100/month

Everything in Pro plus 5x or 20x more usage than Pro, higher output limits, early access to advanced features, priority access at high traffic, and $100–$200/month in Claude Platform API credits.

API usage (Haiku 5.5)

Input $0.10/$0.50 per 1M tokens; Output $0.50/$2.50 per 1M t

Pay-per-token pricing on the Claude Platform and partner clouds; cache reads $0.01/$0.05 and cache writes $0.125/$0.625 per million tokens.

FAQ

What is Claude Haiku 5.5?

Anthropic's fastest and most capable small model, built for high-volume, cost-sensitive tasks like summarization, compaction, classification, database queries, subagent work, live customer support, and browser use.

How much does Haiku 5.5 cost?

Per million tokens: $0.10 input and $0.50 output for prompts up to 100k tokens, rising to $0.50 input and $2.50 output above 100k. Cache reads cost $0.01/$0.05 and cache writes $0.125/$0.625. On average it costs around 75% less to run than Haiku 4.5.

Where is it available?

It is available now on all platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure. On the Claude Platform, developers use the model ID claude-haiku-5-5, and a migration guide covers upgrading from Haiku 4.5.

What is the effort setting?

Claude Haiku 5.5