Claude Haiku 5.5
Anthropic's fastest, lowest-cost model for high-volume work such as classification, extraction, routing, summaries, and sub-agent tasks. Text and image input, 1M context, 128K output including thinking, and medium default effort. Adaptive thinking is on by default; thinking can be disabled at high effort or below. Prompts over 100K tokens are billed at 5x rates.
Specifications
| Provider | Anthropic |
| Category | Text |
| Context window | 1M tokens |
Pricing (Starter tier)
Higher tiers get volume discounts. See tiers
Quick Start
Parameters
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
prompt |
string | required | — | The input text prompt |
system_prompt |
string | optional | — | System instructions |
image_url |
string | optional | — | Image URL to analyze |
max_tokens |
number | optional | 128000 |
Maximum output tokens including thinking, up to 128000. Leave room for thinking on short-answer tasks. |
effort |
string | optional | medium |
low, medium, high, xhigh, or max. Use low for simple high-volume tasks. |
thinking |
object | optional | — | Omit or use type adaptive, optionally with display omitted or summarized. {type: disabled} is accepted at effort high or below. Manual budgets and between_tools are unsupported. |
tools |
array | optional | — | Built-in web_search, web_fetch, and code_execution. Code execution requires web_search or web_fetch version 20260209 or newer alongside it. Queue sessions do not support custom function tools. |
Availability
Claude Haiku 5.5 is available in the GenX Pro App and via the GenX Router API.