Anthropic Releases Claude Haiku 5.5 and Reduces Sonnet 5.5 Caching Costs
Frontier AI models · AI assistants · Enterprise AI
Anthropic introduced Claude Haiku 5.5, a small model designed for high-volume, cost-sensitive, and latency-critical workloads. Anthropic reported an average 75% operating cost reduction compared to Haiku 4.5. Alongside the release, the company halved cache read pricing for Claude Sonnet 5.5, launched monthly API credits for qualifying subscribers, and introduced an adjustable effort setting for Haiku 5.5.
Anthropic announced the launch of Claude Haiku 5.5, positioning it as its fastest and most capable small model to date for repetitive and speed-sensitive tasks such as summaries, database queries, classification, and browser use. The company reported that running Haiku 5.5 costs approximately 75% less on average than Haiku 4.5, with input tokens priced at $0.10 per million for prompts up to 100,000 tokens and $0.50 per million thereafter. Output tokens are set at $0.50 and $2.50 per million, respectively. In internal benchmarks, Haiku 5.5 achieved scores including 72.4% on an offline subset of OSWorld 2.1, 39.2% on Terminal-Bench 4.0, and 45.9% on Humanity's Last Exam without tools. It is also the first model in the Haiku family to support an adjustable effort setting, allowing users to balance intelligence and execution cost. Concurrently, Anthropic halved the price of Claude Sonnet 5.5 cache reads to $0.10 per million tokens, stating this lowers total agentic operational costs by around 20%. Anthropic also introduced monthly API credits for Max ($100), Ultra ($200), and Team (up to $500 pooled) tier subscribers, and added beta computer and browser use capabilities to its Python and TypeScript SDKs. Haiku 5.5 is available on the Claude Platform as well as AWS, Google Cloud, and Microsoft Azure.
Key takeaways
- Anthropic launched Claude Haiku 5.5, featuring an adjustable effort setting and benchmark scores of 72.4% on OSWorld 2.1 offline subset and 39.2% on Terminal-Bench 4.0.
- Running Haiku 5.5 costs an estimated 75% less on average than Haiku 4.5, with input tokens starting at $0.10 per million tokens and output tokens at $0.50 per million for prompts under 100,000 tokens.
- Cache read pricing for Claude Sonnet 5.5 was halved to $0.10 per million tokens, reducing costs for agentic tasks by approximately 20% according to Anthropic.
- Anthropic added monthly API credits for Max, Ultra, and Team subscribers, alongside beta support for computer and browser use in its Python and TypeScript SDKs.
Source：Anthropic Read the original ↗
Prepared, checked and automatically published by AI; not individually reviewed by a human.