Anthropic debuts Claude Opus 5.5 with Fable 5.1-level performance and price cuts, introduces multi-agent swarms

Anthropic announced Claude Opus 5.5 as the first model in a new Claude 5.5 family, stating it performs at the level of Claude Fable 5.1 for most tasks and costs 40% less to run than Opus 5.

The company highlighted speed gains of about 30% and improvements in writing quality, noting the model places the most important information up front and follows user-specified writing rules, which Anthropic framed as addressing common feedback on Opus 5.

Anthropic also reported initial work on large multi-agent swarms, scaling up to 100 parallel agents in its system card, signaling a strategic emphasis on agentic workflows.

Pricing for Opus 5.5 was reduced 20%, from $5/$25 to $4/$20 per 1M input and output tokens, though Anthropic acknowledged higher token usage—especially in coding—can offset some of the savings at higher effort levels.

Independent analysis by Artificial Analysis estimated that at maximum effort, Opus 5.5’s per-task cost rises to $5.98 versus $5.86 for Opus 5 due to increased token usage, with the 20% price cut offsetting part of the increase.

Opus 5.5 is now the default model in Claude Code and the Claude app, including Cowork, with Anthropic describing default effort as "comparable to Fable 5.1 on intelligence but faster."

Anthropic stated that Sonnet 5.5 and Haiku 5.5 will follow in the coming weeks, contradicting earlier speculation that Haiku was discontinued.

The model introduces Fable 5.1-class safeguards on cyber, bio, and frontier LLM development, with Anthropic noting that flagged requests fall back to another model and that the company is working to reduce incorrect flags.