Guides · Models
Claude Haiku 5.5: price, context window and specs
| Developer | Anthropic |
|---|---|
| Released | 7 October 2026 |
| Model ID | claude-haiku-5-5 |
| Context window | 1M tokens (128K max output) |
| API price, prompts up to 100K tokens | $0.10 input / $0.50 output per 1M |
| API price, prompts over 100K tokens | $0.50 input / $2.50 output per 1M |
| Cache reads | $0.01 (up to 100K) / $0.05 (over 100K) per 1M |
| Default effort | medium |
| Claude Code | Default Haiku model on the Anthropic API since v2.1.293 |
Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model we’ve ever released.
What Anthropic says it's for
Anthropic's announcement pitches Haiku 5.5 at quick, repetitive work like summaries, compactions, database queries and classification, and as a subagent next to Opus 5.5 and Sonnet 5.5 on coding work. It says Haiku 5.5 costs around 75% less to run than Haiku 4.5 on average. Anthropic also says Sonnet 5.5 and Opus 5.5 remain the better choice for complex agentic coding. Haiku 5.5 uses the newer tokenizer, so the same text counts as roughly 30% more tokens than on Haiku 4.5.
How prompt-length pricing works
- Haiku 5.5 is the only current Claude model priced by prompt length. A request whose prompt is over 100,000 tokens pays the higher rates for the whole request.
- The prompt length counts all input tokens, including cache reads and cache writes, per Anthropic's pricing page.
- Batch API: 50% off, so $0.05 / $0.25 for short prompts.
- Prompt caching: 5-minute writes $0.125, 1-hour writes $0.20, cache reads $0.01 for prompts up to 100K tokens.
Moving code from Haiku 4.5
Per the release notes, manual extended thinking (budget_tokens), the sampling parameters temperature, top_p and top_k, and assistant message prefill can each return a 400 error on Haiku 5.5, and so does the computer_20250124 tool. Adaptive thinking is on by default and its tokens are billed as output. Haiku 5.5 supports all five effort levels, from low to max.
Using Haiku 5.5 in Claude Code
- Claude Code v2.1.293 (7 Oct 2026) made claude-haiku-5-5 the default Haiku model on the Anthropic API.
- The haiku alias resolves to Haiku 5.5 on the Anthropic API and to Haiku 4.5 on Claude Platform on AWS, Bedrock, Google Cloud and Microsoft Foundry, per the model configuration docs.
- Anthropic's Claude Code guide suggests Haiku for quick mechanical work such as renames, log lines, regex explanations and boilerplate.
Independent benchmarks
Artificial Analysis reported a score of 43 on its Intelligence Index (13 Oct 2026). Anthropic's own benchmark table is on its announcement page, and results from open datasets are on our benchmarks page.
Questions
- When does the higher Haiku 5.5 price apply?
- When a request's prompt is over 100,000 tokens. The whole request is then billed at $0.50 / $2.50 per million tokens, even if part of the prompt is a cache hit. Each request is priced on its own.
- Can I swap Haiku 4.5 for Haiku 5.5 without code changes?
- Not always. budget_tokens thinking, custom temperature, top_p or top_k, and assistant prefill can return 400 errors on Haiku 5.5. Test before you change the model ID.
- Does Haiku 5.5 support effort levels?
- Yes. It supports low, medium, high, xhigh and max, and defaults to medium. Anthropic suggests low for chat, short tool tasks and simple high-volume requests.
Sources
- Claude Haiku 5.5 model page · Anthropic Docs
- Introducing Claude Haiku 5.5 · Anthropic ·
- Introducing Claude Haiku 5.5 (X) · X · @claudeai ·
- Claude API pricing · Anthropic Docs
- Claude Developer Platform release notes · Anthropic Docs
- Effort · Anthropic Docs
- Claude Code model configuration · Claude Code Docs
- Models, usage, and limits in Claude Code · Claude Help Center
- Claude Code v2.1.293 · GitHub · anthropics/claude-code ·
- Artificial Analysis: Claude Haiku 5.5 model page · Artificial Analysis
Last checked . Facts change; the sources above are authoritative. See how we verify or report a correction.
