Breaking

Anthropic Releases Claude Haiku 5.5 as Lower-Cost Small Model for Coding and Computer Use

October 8, 2026 at 01:47 EDT

Foundation Models
AI Agents
Software Dev & Coding

Anthropic released Claude Haiku 5.5 on October 7, positioning it as its cheapest, fastest and most capable small Claude model for high-volume AI workloads. The model is aimed at classification, extraction, routing, summarization, compaction, subagent work, live support, browser use and computer-use tasks, with the API model ID claude-haiku-5-5 and availability across Claude API, Amazon Bedrock, Google Cloud, Microsoft Foundry and Claude Platform on AWS.

Oct. 7, 2026 · Anthropic

Claude Haiku 5.5 moves small-model work downmarket

Anthropic's new small Claude model targets high-volume coding support, computer use, routing and document tasks with lower prices and stronger benchmark scores than Haiku 4.5.

Price drop, drawn to scale

$0.10

Haiku 5.5 input per MTok

$1.00

Haiku 4.5 input per MTok

For prompts up to 100,000 tokens, the published input price is one-tenth of Haiku 4.5.

context window

128K

max output

72.4%

OSWorld score

75%

average cost reduction

Benchmark position

OSWorld 2.1

72.4%

Above Haiku 4.5 and GPT-6 Luna; below Sonnet 5.5.

Terminal-Bench 4.0

39.2%

Improved over Haiku 4.5, but complex coding still favors Sonnet and Opus.

Where it fits

High-volume classification, summaries, compaction, subagents, live support and browser or computer-use tasks.

What to watch

Migration requires prompt recounting, adaptive thinking changes, sampling-parameter cleanup and updated computer-use tooling.

The strategic shift

Haiku 5.5 does not erase the need for larger Claude models. It gives developers a cheaper, faster worker for repeatable tasks that can run beside Sonnet or Opus in larger agent systems.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.