Anthropic releases Claude Opus 4.5

Background

Claude Opus 4.5 followed Claude Opus 4.1 in the Opus line. Anthropic offers Claude through the Claude apps, Claude Code, the Claude API and cloud platforms, and developers pay per million input and output tokens, a token being the unit of text a model reads and writes.^[1]

What happened

On 24 November 2025 Anthropic released Claude Opus 4.5 at $5 per million input tokens and $25 per million output tokens. That was a third of the previous Opus price.^[1]

Anthropic said the model was state-of-the-art on SWE-bench Verified, a coding benchmark. It scored higher than any human candidate on Anthropic's take-home engineering exam within the two-hour limit.^[1]

Key details

A new effort parameter let developers trade capability for speed and cost.^[1]

Claude Code gained Plan Mode and multiple parallel sessions in the desktop app.^[1]

Long conversations in the Claude apps no longer ended at a limit, because earlier context was summarised automatically.^[1]

Claude for Chrome became available in beta to all Max subscribers, adding scheduled tasks and multi-tab workflows.^[1][2]

Opus-specific usage caps were removed for users with Opus 4.5 access.^[1]

An effort parameter is a setting in the API that tells the model how much work to put into an answer. A lower setting favours speed and cost, and a higher setting favours thoroughness, so developers can choose the trade-off that suits each request.

Impact

The release lowered the price of the Opus tier and expanded what users could do in Claude Code and the Claude apps, in areas such as long conversations and browser work. The effort parameter gave developers a direct control over how much capability to pay for in each request.^[1] Summarised earlier context keeps long conversations running, but the exact wording from early in a conversation may no longer be kept in full.