AI Agent Hub
Back to models
Anthropic: Claude Haiku Latest logo

Anthropic: Claude Haiku Latest

Closed Source ~anthropic Released 2026-10-07
-- 1M context Proprietary

About this model

Claude Haiku Latest (Claude Haiku 5.5) is Anthropic’s current small, fast Claude model, released in October 2026 as the default Haiku tier on the Claude API (claude-haiku-5-5). It targets high-volume, latency-sensitive workloads such as classification, routing, extraction, summarization, compaction, and lightweight subagent roles alongside larger Sonnet and Opus models. Adaptive thinking with an adjustable effort parameter lets developers trade quality against speed and cost without changing model IDs.

The model accepts text and image inputs and returns text, with a one-million-token context window and up to 128K output tokens. It shares a newer tokenizer with later Claude releases (roughly 30% more tokens than Haiku 4.5 for the same text) and supports computer-use and browser-use workflows in beta via updated SDKs. Pricing is sharply lower than Haiku 4.5, with tiered input and output rates for prompts above 100K tokens.

On Anthropic’s pre-deployment evaluations at max effort, Haiku 5.5 substantially outperforms Haiku 4.5 on agentic coding, computer use, multilingual knowledge (GMMLU), and Humanity’s Last Exam, while remaining below Sonnet 5.5 on the hardest long-horizon coding and terminal tasks. It is proprietary and available only through Anthropic’s API and cloud partners (AWS Bedrock, Google Cloud, Microsoft Foundry), not as downloadable weights.

Benchmark Scores

HLE
57.4
MMLU
87.8
GDPval-AA
1620.0
LiveBench
72.1
SWE-Bench-Pro
64.8
SWE-Bench-Multilingual
83.7

Technical Specs

  • Architecture: Transformer
  • Context Window: 1,000,000 tokens
  • Input Modalities: text, image

Hardware Requirements

  • Compute: API only

Pricing

Input Output Currency
0.10 / 1M tokens 0.50 / 1M tokens USD