Claude Haiku 5.5Latest

For high-volume, latency-sensitive tasks such as classification, extraction, and routing

Context window: 1Mtokens
Max output: 128Ktokens
Input pricing: From $0.10/ MTok
Output pricing: From $0.50/ MTok

概述

Claude Haiku 5.5 专为高吞吐量、对延迟敏感的工作而构建，例如分类、路由、提取和子智能体任务。它支持带有 effort 参数的 adaptive thinking（自适应思考）、1M 令牌的上下文窗口，以及最多 128k 输出令牌。它使用与 Claude 4.7 及更高版本模型相同的较新 tokenizer（分词器），因此相同文本的令牌数比在 Claude Haiku 4.5 上多约 30%。它的 thinking block（思考块）仅在生成它们的账户或与该账户关联的账户中有效。

有关代码变更，请参阅迁移指南。有关模型 ID、定价和限制，请参阅 Claude Haiku 5.5 概述。有关提示指导，请参阅为 Claude Haiku 5.5 编写提示。

Claude Haiku 5.5 的新功能

对比情况

Model	Context	Max output	Price / MTok	Latency	Thinking	Default effort	Knowledge cutoff
Claude Fable 5.1	1M	128K	$10 / $50	Slower	Adaptive (always on)	`high`	Jun 2026
Claude Opus 5.5	1M	128K	$4 / $20	Moderate	Adaptive (always on)	`medium`	Jun 2026
Claude Sonnet 5.5	1M	128K	$2 / $10	Fast	Adaptive	`high`	Jun 2026
Claude Haiku 5.5This model	1M	128K	From $0.10 / $0.50	Fastest	Adaptive	`medium`	Jun 2026

规格

Model IDs

Claude API
Amazon Bedrock
Google Cloud
Microsoft Foundry
Claude Platform on AWS

Pricing

Input: $0.10 / MTok for prompts up to 100,000 tokens$0.50 / MTok for prompts over 100,000 tokens
Output: $0.50 / MTok for prompts up to 100,000 tokens$2.50 / MTok for prompts over 100,000 tokens
5m cache write: $0.125 / MTok for prompts up to 100,000 tokens$0.625 / MTok for prompts over 100,000 tokens
1h cache write: $0.20 / MTok for prompts up to 100,000 tokens$1 / MTok for prompts over 100,000 tokens
Cache read: $0.01 / MTok for prompts up to 100,000 tokens$0.05 / MTok for prompts over 100,000 tokens
Batch API: 50% discount on input and output

Full price list

Capabilities

Context window: 1M tokens
Max output: 128K tokens
Max output (Batch API, beta): 300K tokens
Thinking: Adaptive
Default effort: medium
Comparative latency: Fastest
Input → output: Text and images → text
Reliable knowledge cutoff: Jun 2026
Training data cutoff: Jun 2026

Availability

Status: Active (latest)
Released: October 7, 2026
Retirement: Not sooner than October 7, 2027
Platforms: Claude APIAmazon Bedrock Google Cloud Microsoft Foundry Claude Platform on AWS

须知事项

"Adaptive thinking"（自适应思考）默认开启。使用 effort 参数控制思考深度。
请省略 temperature、top_p 和 top_k，因为其中任何一个设置为非默认值都会返回 400 错误。
在 Message Batches API 上，Claude Haiku 5.5 配合 output-300k-2026-03-24 beta 标头可支持最多 300k 输出令牌。
使用 Models API 以编程方式查询限制和功能。