Stories
Anthropic Unveils Claude Haiku 5.5 with Enhanced Speed, Lower Cost and New Effort Controls
- •Anthropic has launched Claude Haiku 5.5, touted as its cheapest, fastest, and most capable smaller AI model yet designed for quick workloads and sub-agent tasks.
- •The new model introduces an effort setting, allowing users to choose between optimizing for cost or intelligence based on their specific needs.
- •Haiku 5.5 delivers major safety improvements over its predecessor, featuring stricter cybersecurity and biology safeguards to prevent misuse.
- •Anthropic has halved the price of cache reads for Sonnet 5.5 and is offering new monthly API credits for Claude Max and Team subscribers.
- •The model is available now across platforms including Amazon Web Services, Google Cloud, and Microsoft Azure.
Key Questions Answered
Anthropic launched Claude Haiku 5.5 as its cheapest, fastest, and most capable smaller model designed for quick repetitive workloads like summaries and database queries, featuring input token pricing starting at $0.10 per million tokens for prompts up to 100K. It also introduces a new effort setting to balance cost or intelligence, alongside stricter cybersecurity and biology safeguards.
Claude Haiku 5.5 is currently available across all major platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure, allowing developers to get started immediately on the Claude platform.
Haiku 5.5 demonstrates major improvements over Haiku 4.5 by showing fewer instances of misaligned behavior, lower willingness to cooperate with misuse, and stricter cybersecurity safeguards that permit a broader range of defensive tasks while blocking cyberattack techniques.
