Breaking

Anthropic Launches Claude Opus 5.5 With Lower Pricing and Longer Claude Code Limits

September 22, 2026 at 12:44 EDT

Foundation Models
Software Dev & Coding
AI Agents

Anthropic launched Claude Opus 5.5 on September 22, 2026, positioning it as the first model in its Claude 5.5 family and a cheaper, faster successor to Opus 5 for agentic coding and knowledge work. The company says the model performs at about the level of Claude Fable 5.1 on most work while costing 40% less per typical task than Opus 5.

Sep. 22, 2026 | Anthropic

Claude Opus 5.5 shifts the agent-cost curve

The new Opus release targets long coding and knowledge-work agents with Fable-level performance, lower token prices and more Claude Code headroom.

40%

lower cost per typical task vs Opus 5

>30%

faster output generation than Opus 5

token context window on the Claude Platform

20%

higher five-hour Claude Code limits

Cache reads get the steepest price cut

Cost per 1M cache-read tokens, shown with one block equal to $0.10.

$0.20Opus 5.5

$0.50Opus 5

60% lower cache-read priceA meaningful cut for coding agents that repeatedly reuse large project context.

Benchmark signal

Terminal-Bench 4.0

66.4%Opus 5.5 top-line score

FrontierCode v1.1

54.4%merge-oriented coding tasks

GDPval-AA v2.1

1846Elo on knowledge work

What improves

Lower token and cache costs for long agent runs
Clearer communication over long coding sessions
Higher Claude Code limits and a banked reset

What remains constrained

Cyber and biology work routes through verification and safeguards
Thinking mode is always on for Opus 5.5
Evaluation gaps remain a stated safety concern

Continue reading

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