Breaking

Anthropic Launches Claude Opus 5.5, Citing 40% Lower Typical Cost Than Opus 5

September 22, 2026 at 15:49 EDT

Foundation Models
AI Agents
Software Dev & Coding

Anthropic released Claude Opus 5.5 on September 22, 2026, positioning it as the first model in its Claude 5.5 family for long-running agentic coding and knowledge work. The company says the model performs at roughly Claude Fable 5.1 level on most work while costing 40% less to run than Claude Opus 5 on typical token-billed workloads.

Sep. 22, 2026 · Anthropic

Claude Opus 5.5 shifts the Opus equation

A new Opus model targets long-running agents with lower token prices, fewer steps per task, and stronger coding scores while keeping stricter safeguards in place.

40%

lower typical token-billed workload cost than Opus 5

$4 / $20

input and output pricing per million tokens

66.4%

Terminal-Bench 4.0 score reported for Opus 5.5

documented context window in tokens

The cost index drops from 100 to 60

Vertical blocks show the claimed typical workload cost, scaled to Opus 5 as 100.

Opus 5

100 index

Opus 5.5

60 index

What changes in the agent loop

Lower token price

→

Fewer steps

→

Faster output

→

Longer-running agents

What builders like

Clearer, shorter answers
Lower token use in long coding runs
Better fit for code review and migrations

What still needs care

Usage limits remain a planning constraint
Always-on thinking changes integrations
High-risk cyber and biology work is gated

A faster Opus with tighter guardrails

The model arrives with stronger reported alignment scores, containment improvements, and Fable-class safeguards for sensitive cyber, biology, and distillation use cases.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.