Claude Opus 5.5
Anthropic · Multimodal model
Claude Opus 5.5 is Anthropic's main frontier model, released on September 22, 2026, and the model Anthropic suggests starting with for most work. It performs close to Fable 5.1 on most tasks for less than half the price per token.
Current
ReleasedSep 22, 2026
Context window1M tokens (128K max output)
Input → outputtext, image → text
API price$4 / input MTok, $20 / output MTok
Consumer accessClaude Pro, Max, Team and Enterprise; not on Free
Best for
Long-running agentic coding and knowledge work; Anthropic's recommended default for most workloads.
Overview
Claude Opus 5.5 is Anthropic's main frontier model, released on September 22, 2026, and the model Anthropic suggests starting with for most work. It performs close to Fable 5.1 on most tasks for less than half the price per token.
Strengths and limits
- Leads Anthropic's published table on Terminal-Bench 4.0, CursorBench 4.0 and GDPval-AA
- About Fable 5.1 level on most tasks at less than half the token price
- Fast mode (up to 2.5x speed) in Claude Code and the Claude Platform
- Thinking cannot be turned off
- Most cybersecurity tasks are rerouted to Opus 4.8 under safeguards
Published benchmark scores
Terminal-Bench 4.066.4%
Source: www.anthropic.com
Humanity's Last Exam (with tools)67.7%
Source: www.anthropic.com
OSWorld 2.1 (partial)81.8%
Source: www.anthropic.com
GDPval-AA v2.11846
Source: www.anthropic.com
Scores are as published by the source; settings (reasoning effort, tools, attempts) differ between labs, so compare with care. How to read benchmarks →