Anthropic expands Claude lineup with faster, lower-cost models

Anthropic introduced Claude Haiku 5.5 on Oct 7, 2026, describing it as its fastest and most capable small model. It targets high-volume work such as summarization, subagents, and browser use. The release follows updates to the Sonnet and Opus families focused on stronger performance and lower operating costs.

Claude Sonnet 5.5 runs 30%+ faster and costs up to 30% less for most work than Claude Sonnet 5, according to Anthropic. The company says Claude Opus 5.5 improves agentic coding and knowledge work while costing 40% less to run than Opus 5 on typical workloads. Anthropic also introduced Claude Mythos and Fable 5.1, positioning their coding, knowledge-work, and research capabilities as a step toward AI contributions to scientific progress.

Alongside model development, Anthropic has committed to preserving the weights of all publicly released models and those deployed for significant internal use. Claude Opus 3 remains available after retirement to paid users on claude.ai and by request through the API. The company is also proposing a framework for scoring jailbreak severity with Amazon, Microsoft, Google, and other Glasswing partners.