Anthropic unveils new Claude 5.5 AI models
Anthropic has introduced `Claude Fable 5.2`, `Claude Opus 5.5` and `Claude Sonnet 5.5` for programming, science, finance and knowledge work. The company says Opus 5.5 costs 40% less than Opus 5 for typical workloads, while Sonnet 5.5 is more than 30% faster and costs up to 30% less for most workloads.

Anthropic has introduced a new generation of Claude models aimed at programming, finance, science and knowledge work. The company highlights Claude Fable 5.2, Claude Opus 5.5 and Claude Sonnet 5.5 as its most capable options for demanding technical tasks.
The announcement appears on Anthropic’s homepage. The company is a public benefit corporation that says it works to expand the benefits of AI and reduce its risks. It places safety at the center of development, rather than treating it as a feature added at the end.
Three models for different types of work
Claude Fable 5.2 is designed for complex tasks in programming, finance and science. Anthropic describes it as a model built for high-level technical work, although the published information does not provide details on testing, availability dates or pricing.
Claude Opus 5.5 focuses on agentic programming and knowledge work. Agentic programming allows AI to break a task into steps, use tools and make progress with fewer ongoing instructions from a person.
According to Anthropic, Opus 5.5 costs 40% less than Opus 5 for typical workloads. The exact comparison depends on the type of task and how the model is used, so it does not mean that every query will cost exactly 40% less.
Claude Sonnet 5.5 is the family’s more accessible update. The company says it:
- Works more than 30% faster than Claude Sonnet 5.
- Costs up to 30% less for most workloads.
- Is designed for general tasks that need a balance of speed, capability and cost.
What this means for you
If you use Claude to write code, analyze documents or conduct research, these changes could mean faster responses and lower costs for repetitive work. In a company, for example, Sonnet 5.5 could handle reviews of large volumes of text, while Opus 5.5 could take on longer, more autonomous programming tasks.
For users, important information is still missing: how much each model will cost, who will be able to access them and how they perform in independent tests. The published figures are Anthropic’s claims and come with conditions such as “typical workloads” or “for most workloads.”
The main signal is that Anthropic is competing on two fronts at once: more capable models for technical tasks and faster, cheaper models for everyday use. What to watch now is whether those improvements hold up beyond the company’s chosen scenarios and when the models actually reach users.

