“I want walls of glass to fall asleep under the stars, a deck cantilevered over still water to drink my coffee above the fish, and a little stone tower where a telescope pokes through the dome glass — books, sky, and water in one quiet place, with the forest at my back so I’m never quite alone.”
Anthropic · Claude
Claude Sonnet 5
Anthropic’s most agentic Sonnet yet — near Opus-class capability at a fraction of the price.
Yes — Claude Sonnet 5 is a real Anthropic model.
Released June 2026 by Anthropic, Claude Sonnet 5 is part of the Claude family. It's available through Anthropic's API, and we run it on PromptFrenzy — the examples below are real, unedited generations it produced.
Maker
AnthropicReleased
June 2026
Input price
$2 / 1M
Output price
$10 / 1M
What is Claude Sonnet 5?
Claude Sonnet 5 launched on 30 June 2026 as the successor to Sonnet 4.6 in Anthropic’s mid-tier line. Anthropic positions it as their most agentic Sonnet to date — planning, tool use, and autonomous multi-step work that approaches the capability of the larger Opus 4.8 at a much lower price. It also reports lower hallucination and sycophancy than Sonnet 4.6, and stronger resistance to prompt-injection attacks.
Strengths
- ✓Agentic work — planning, tool use, and multi-step coding near Opus-class.
- ✓Strong capability-per-dollar; far cheaper than the flagship tier.
- ✓Lower hallucination and sycophancy, better prompt-injection resistance than Sonnet 4.6.
Trade-offs
- ✗As a mid-tier model it trades some peak depth on the hardest reasoning tasks for speed and cost.
- ✗Brand-new (June 2026) — long-horizon reliability is still being established in the wild.
What Claude Sonnet 5 actually produces
Real, unedited outputs from Claude Sonnet 5 in our model showdowns — same brief, same constraints as every other model, one shot each.
See Claude Sonnet 5 compared head-to-head
Compare every model side-by-side
See how Claude Sonnet 5 stacks up against the other frontier models on the same brief.
Browse all model showdowns →