Breaking

Anthropic Releases Claude Opus 5.5 as Agent Arena Opens Live Voting on the Model

September 22, 2026 at 13:48 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 the Claude 5.5 family and a lower-cost upgrade for long-running agents, coding and knowledge work. Agent Arena has also added Claude Opus 5.5 for community voting, putting the new model into a live evaluation setting focused on real-world agentic tasks.

September 22, 2026 · Anthropic

Claude Opus 5.5 enters the agent race

The new Opus model targets long-running coding agents and knowledge work with lower token prices, faster output and live community evaluation in Agent Arena.

40%

Lower typical workload cost than Opus 5

Token context window

128K

Maximum output tokens

66.4%

Terminal-Bench 4.0 score

Cost drops are built into the unit economics

Input, output and cache-read prices all move lower than Opus 5, with the steepest proportional cut on cache reads.

Opus 5.5 input

Opus 5 input

Cache reads fall from $0.50 to $0.20 per million tokens, a 60% reduction for repeated long-context work.

What improves

Lower token and cache-read prices
More than 30% faster output than Opus 5
Strong reported gains in coding and knowledge work

What changes

Adaptive thinking is always on
Forced tool use is not supported
Older computer-use tooling needs migration

Agent workflow

Plan

→

Use tools

→

Check work

→

Report clearly

Continue reading

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