Quick Answer
Claude Haiku 5.5 is Anthropic’s newest small, fast and low-cost AI model, released October 7, 2026. It is designed for high-volume tasks such as summarization, document extraction, classification, routing, customer support and AI-agent sub-tasks. It supports a 1-million-token context window, up to 128,000 output tokens, adaptive thinking and adjustable effort. API prices start at $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens. It is available on Claude.ai, Claude’s API and major cloud platforms.
What Is Claude Haiku 5.5?
Claude Haiku 5.5 is a compact language model from Anthropic, the company behind Claude. Haiku models prioritize speed and cost efficiency, while the Sonnet and Opus lines generally target more demanding workloads. The new model is intended for situations where an organization must process many requests without paying the cost of a larger frontier model every time.
Anthropic describes Haiku 5.5 as its fastest, cheapest and most capable small model so far. Typical applications include extracting facts from documents, classifying support requests, answering short questions, routing tasks to specialized agents and supporting larger models in software development.
Claude Haiku 5.5 Release Date and Availability
Anthropic officially announced Claude Haiku 5.5 on October 7, 2026. It is available to Free, Pro, Max, Team and Enterprise users on Claude.ai through supported web and mobile experiences, as well as Claude Code. Developers can access it via the Claude API using the model ID claude-haiku-5-5. Anthropic also lists availability through Amazon Bedrock, Google Cloud and Microsoft Foundry. Access can vary by product and account settings.
Key Claude Haiku 5.5 Features
1. Speed for real-time and high-volume work
Anthropic positions Haiku 5.5 as its fastest standard-speed model. This is valuable for customer-service chat, content tagging, rapid lookups and applications where response latency matters. Speed varies with prompt length, output size and service conditions.
2. One-million-token context window
The model supports a 1M-token context window, allowing large documents or substantial conversation history to be provided in one request, subject to API and product limits. A large context window does not guarantee perfect recall of every detail.
3. Up to 128K output tokens
Anthropic documents a maximum output of 128,000 tokens for ordinary API use. Batch API users may have access to a larger beta output allowance under specific conditions. Real applications should set appropriate output limits to manage cost and latency.
4. Adaptive thinking and adjustable effort
Haiku 5.5 is the first Haiku-class model with an adjustable effort setting. Developers can trade reasoning effort against cost and latency according to the task. Anthropic lists medium as its default effort.
5. Image understanding and tool use
Claude Haiku 5.5 accepts text and image inputs and generates text. It supports tool use, making it suitable for structured workflows, extraction and agent applications. It is not an image-generation model.
6. Computer-use and browser-use workflows
Anthropic highlights computer-use tasks and announced beta support in its Python and TypeScript SDKs. Agents can use such capabilities only where the surrounding product, tools and permissions support them. Human review remains important for sensitive actions.
Claude Haiku 5.5 Pricing Explained
Anthropic prices Haiku 5.5 according to the length of the request’s prompt. The following are Claude API prices per one million tokens, not Claude subscription fees.
| Usage | Prompt up to 100K tokens | Prompt over 100K tokens |
|---|---|---|
| Input tokens | $0.10 | $0.50 |
| Output tokens | $0.50 | $2.50 |
| Cache reads | $0.01 | $0.05 |
| 5-minute cache writes | $0.125 | $0.625 |
For example, at the lower prompt-length tier, a request using 10,000 input tokens and 2,000 output tokens would cost approximately $0.002 before any other charges or discounts: $0.001 for input and $0.001 for output. Real bills depend on actual token usage and pricing rules. Anthropic also offers a 50% discount on eligible Batch API input and output usage.
Anthropic says average operating cost is approximately 75% lower than Haiku 4.5, accounting for request distribution and tokenization differences. That is a company estimate, not a guaranteed saving for every application.
Claude Haiku 5.5 vs Haiku 4.5 vs Sonnet 5.5
| Feature | Haiku 4.5 | Haiku 5.5 | Sonnet 5.5 |
|---|---|---|---|
| Primary focus | Earlier compact model | Fast, economical high-volume tasks | More demanding general and agentic tasks |
| Input price / 1M tokens | $1 | From $0.10 | $2 |
| Output price / 1M tokens | $5 | From $0.50 | $10 |
| Adjustable effort | Not the new Haiku 5.5 feature | Yes | Yes |
| Best fit | Legacy workloads | Volume, speed and budget | More complex tasks |
These are published list prices for the respective model tiers, not a claim that every task is cheaper or better with Haiku 5.5. Anthropic’s own benchmarks show that Sonnet 5.5 remains stronger on several demanding reasoning and agentic coding evaluations.
Performance and Benchmarks
In Anthropic’s published evaluations, Haiku 5.5 scored 72.4% on the OSWorld 2.1 offline subset for computer use, compared with 15.7% for Haiku 4.5. On Humanity’s Last Exam without tools, Anthropic reports 45.9% for Haiku 5.5 versus 10.2% for Haiku 4.5. These are vendor-reported benchmarks under specified evaluation conditions; they should not be interpreted as guaranteed performance on everyday tasks.
For complex agentic coding, Anthropic still recommends its larger Sonnet and Opus models. Haiku 5.5’s advantage is the balance between cost, speed and capability for narrower tasks.
How to Use Claude Haiku 5.5
On Claude.ai
- Visit Claude.ai and sign in.
- Open the model selector, if available in your account.
- Choose Claude Haiku 5.5.
- Enter a prompt such as “Summarize this document in five clear bullet points.”
For developers
Use the Claude API model identifier claude-haiku-5-5 with your existing supported Claude SDK or API integration. Check the official model documentation for limits, SDK requirements and current billing details.
Best Uses for Claude Haiku 5.5
- Customer support: triage and answer routine requests with appropriate safeguards.
- Document processing: extract fields, categorize reports and summarize large text collections.
- AI agents: perform quick research, routing and subagent tasks alongside larger models.
- Business workflows: classify emails, organize tickets and process repeated questions.
- Development: support lightweight code-review, issue-triage and tool-calling steps.
For readers comparing assistants, see our Claude vs ChatGPT guide and the latest GPT-6 overview. You can also explore our ChatGPT Free vs Plus comparison.
Limitations and Things to Know
Haiku 5.5 can still make factual mistakes, misunderstand ambiguous documents and produce incorrect code. Its knowledge cutoff is June 2026, so current information may require search or external tools. Anthropic notes that its newer tokenizer can count the same text as roughly 30% more tokens than Haiku 4.5; evaluate real costs using actual API usage rather than assuming identical token counts. For sensitive or high-impact tasks, use validation, human oversight and appropriate access controls.
Frequently Asked Questions
What is Claude Haiku 5.5?
It is Anthropic’s fast, lower-cost small AI model released on October 7, 2026, for high-volume tasks and lightweight agents.
When was Claude Haiku 5.5 released?
Anthropic announced it on October 7, 2026.
Is Claude Haiku 5.5 free?
Anthropic lists Haiku 5.5 as selectable for Claude.ai Free users, subject to product availability and usage limits. Developer API use is billed separately.
How much does Claude Haiku 5.5 cost?
API pricing starts at $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens; higher prices apply above that threshold.
What is its context window?
Claude Haiku 5.5 supports a 1-million-token context window.
What is the Claude Haiku 5.5 API model ID?
The Claude API model identifier is claude-haiku-5-5.
Is Haiku 5.5 better than Sonnet 5.5?
Haiku 5.5 prioritizes speed and cost, while Sonnet 5.5 is generally better suited to complex tasks. The best model depends on workload and budget.
Final Verdict
Claude Haiku 5.5 is a significant update for anyone building fast, affordable AI workflows. Its low starting API prices, adaptive thinking, long context window and broad availability make it attractive for repeated document tasks, customer support and subagents. For demanding coding and reasoning, evaluate Sonnet or Opus as alternatives. Pricing and product availability may change, so check Anthropic’s official documentation before making deployment decisions.
Editorial update: October 8, 2026. This article is based on Anthropic’s official launch announcement and model documentation.