Claude Sonnet 5

[reasoning]

Anthropic[text]

Most agentic Sonnet yet, near-Opus 4.8 performance at a fraction of the cost (intro pricing $2/$10 through Aug 31)

Input

per 1M tokens

Output

$15

per 1M tokens

Cached

$0.30

per 1M tokens

Context

1M tokens

Max Output

128K tokens

Released

2026-06-30

Features

[optional-reasoning][1m-tokens-context][near-opus-4.8-performance][autonomous-task-execution][self-checking-output][prompt-caching-90%-discount]

Technical Specifications

Benchmarks

ARC-AGI-284.7%

GPQA Diamond96.2%

OSWorld-Verified88.3%

SWE-bench Verified92.4%

The Lab

location: San Francisco, California, USAfounded: 2021backed by: Amazon, Google; public benefit corporation

Founded in 2021 by former OpenAI researchers and headquartered in San Francisco, Anthropic is a public benefit corporation backed by Amazon and Google. The lab pursues safety-first AI research using Constitutional AI to align models with human values. Its Claude family ships in Haiku, Sonnet and Opus tiers, with the Fable tier added in June 2026 as a Mythos-class model released with additional safeguards.

more about Anthropic →

Model ID: claude-sonnet-5