Claude Sonnet 5.5
Anthropic · Multimodal model
Claude Sonnet 5.5 is Anthropic's mid-tier model, released on September 28, 2026, balancing speed and intelligence. It scores close to Opus 5.5 on many benchmarks at half the price and is the model behind the free Claude plan.
Current
ReleasedSep 28, 2026
Context window1M tokens (128K max output)
Input → outputtext, image → text
API price$2 / input MTok, $10 / output MTok
Consumer accessClaude Free, Pro, Max, Team and Enterprise
Best for
Everyday coding, bug fixing and document/spreadsheet work where speed and cost matter more than peak judgment.
Overview
Claude Sonnet 5.5 is Anthropic's mid-tier model, released on September 28, 2026, balancing speed and intelligence. It scores close to Opus 5.5 on many benchmarks at half the price and is the model behind the free Claude plan.
Strengths and limits
- Near-Opus 5.5 scores on several agentic benchmarks at half the token price
- 30%+ faster than Sonnet 5 and uses fewer tokens per task
- Available on the free Claude plan
- Anthropic says Opus 5.5 is clearly stronger on complex, open-ended work
- Higher-risk security requests are routed back to Sonnet 5
Published benchmark scores
Terminal-Bench 4.070.6%
Source: www.anthropic.com
OSWorld 2.1 (partial)80.1%
Source: www.anthropic.com
Humanity's Last Exam (with tools)64.5%
Source: www.anthropic.com
GDPval-AA v2.11844
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 →