ARTIFICE DAILY AI + CYBER SECURITY BRIEFING

Artificial Intelligence

Anthropic launches Claude Haiku 5.5

Anthropic said Claude Haiku 5.5 was its first Haiku model with adjustable effort, lower priced than its predecessor, and available across major cloud platforms for repetitive and speed-sensitive tasks.

AnthropicAI

Anthropic introduced Claude Haiku 5.5 as a model for high-volume, cost-sensitive work, including summaries, compaction, database queries and classification. It also said the model suited coding subagent work and live customer support because of its speed.

The company said Haiku 5.5 was its first Haiku-class model with an adjustable effort setting. It said average running costs were about 75% lower than for Haiku 4.5, with different pricing depending on token counts.

Anthropic said the model scored better on its alignment evaluations than Haiku 4.5 and showed fewer cases of misaligned behaviour. It said the cybersecurity safeguards were stricter than Haiku 4.5’s, but still allowed more defensive work than some newer models while blocking penetration testing and related attack techniques.

Haiku 5.5 was available on Amazon Web Services, Google Cloud, Microsoft Azure and the Claude Platform. Anthropic also said it had halved cache-read prices for Claude Sonnet 5.5 and would give monthly API credits to Max and Team subscribers.

Named in this story

Companies

Anthropic
announced the model and related pricing changes
Amazon Web Services
was one of the cloud platforms carrying the model
Google Cloud
was one of the cloud platforms carrying the model
Microsoft Azure
was one of the cloud platforms carrying the model

Products and systems

Claude Haiku 5.5
the new Haiku model introduced in the post
Claude Haiku 4.5
the previous Haiku version used for price comparisons
Claude Sonnet 5.5
had its cache-read price cut and was cited in comparisons
Opus 5.5
was named as a partner model for coding work
Claude Sonnet 5
was named in the biology safeguards comparison
Claude Max
its subscribers were offered monthly API credits
Claude Team
its subscribers were offered monthly API credits
Claude Platform
was one place where the model was available

How the source tells it

A promotional product announcement that mixed self-praise with safety reassurance.

  • vendor boosterism superlatives and repeated self-comparison to earlier models
  • novelty the launch is framed through first-time features and new pricing changes
  • reassurance the safety section uses alignment and safeguards language to allay concern