> ## Content Index
> Fetch the complete content index at: https://globalfeed.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# Anthropic releases Claude Haiku 5.5, its first small model with an adjustable effort setting
- URL: https://globalfeed.ai/en/anthropic-releases-claude-haiku-5-5-its-first-small-model-with-an-adjustable-effort-setting/
- Published: 2026-10-07T23:51:04.000Z
- Updated: 2026-10-07T23:51:04.000Z
- Description: Anthropic made Claude Haiku 5.5, the newest member of its smallest model family, available on October 7, 2026. The model is built for high-volume work such as summaries, classification and subagent tasks, and it is the first Haiku with an adjustable effort setting. The same day, Anthropic cut the ca
- Author: GlobalFeed Editor
- Tags: Anthropic, Claude, AI model, small model, x-claudeai, dil-en, elle

**Anthropic** released **Claude Haiku 5.5** on October 7, 2026\. Haiku is the smallest and fastest model line in the Claude family. According to the company's product page, Haiku 5.5 is the cheapest, fastest and most capable small model Anthropic has released so far. It is available in the API as `claude-haiku-5-5` and, according to Anthropic, on all platforms, including Amazon Web Services, Google Cloud and Microsoft Azure.

Anthropic positions Haiku 5.5 for high-volume, cost-sensitive work such as summarizing text, compacting long conversations, database queries and classification. The company says the model works well as a subagent alongside Claude Opus 5.5 and Sonnet 5.5 on coding work, and that it suits speed-sensitive jobs like live customer support and browser use.

## The Haiku line gets an effort setting

Haiku 5.5 is the first Haiku model with an adjustable effort setting. The setting controls how long the model thinks about a task, with five levels from low to max. Developers can use it to decide, task by task, whether to favor cost or quality. In the charts Anthropic shared, both success rates and cost per task rise as effort goes up.

According to the company, Haiku 5.5 is especially strong on tasks under 100,000 tokens, which Anthropic says make up the large majority of requests to its previous Haiku model. Pricing has two tiers based on prompt length, and Anthropic says the model is, on average, considerably cheaper to run than Haiku 4.5\. Current prices are on Anthropic's site.

## Test results published by Anthropic

In Anthropic's table, Haiku 5.5 is well ahead of Haiku 4.5 in coding, computer use and knowledge work. According to the company's results, the model scores 72.4 percent on OSWorld 2.1, a test of operating a computer through the screen, 39.2 percent on the agentic coding test Terminal-Bench 4.0 and 45.9 percent on Humanity's Last Exam without tools. The same table includes OpenAI's GPT-6 Luna; Haiku 5.5 is ahead of Luna on every test where Luna has a score, while Sonnet 5.5 scores above Haiku 5.5 on all of them. The knowledge work measure, GDPval-AA, is run by Artificial Analysis, and the test methods are described in the model's system card.

## Safety settings

Anthropic says Haiku 5.5 shows large improvements over Haiku 4.5 on almost all of its alignment evaluations, with fewer instances of misaligned behavior and less willingness to help with misuse. For cybersecurity, the model's safeguards are stricter than Haiku 4.5's and looser than Sonnet 5.5's; techniques that attackers could also use, such as penetration testing, are still blocked. For biology, the same safeguards as Sonnet 5.5 and Opus 5 apply. Organizations that need broader access in these areas can apply to Anthropic's verification programs.

## Cache reads get cheaper on Sonnet 5.5

The same day, Anthropic halved the price of cache reads for Claude Sonnet 5.5\. Cache reads make up much of the cost of long-running agent work, where the same long context is sent again and again; the company says the cut makes Sonnet 5.5 cheaper to run on most agentic work.

The published test results and cost comparisons come from Anthropic. The product page does not state the model's context window or how it performs in languages other than English, so it is worth testing it on your own examples first. The customer quotes on the page reflect those companies' own experience. For a fuller picture, follow evaluations from independent researchers as well.

Sources: [Anthropic: Claude Haiku 5.5](https://www.anthropic.com/claude-haiku-5-5?ref=globalfeed.ai), [Anthropic: Claude Haiku 5.5 system card](https://www.anthropic.com/claude-haiku-5-5-system-card?ref=globalfeed.ai), [Claude docs: Haiku 5.5 migration guide](https://platform.claude.com/docs/en/models/haiku-5-5/migration-guide?ref=globalfeed.ai).