Term
Claude Sonnet 5
Claude Sonnet 5 is Anthropic's mid-tier model from June 30, 2026 — the most agentic Sonnet yet, with performance close to Opus 4.8 at $2 input and $10 output per 1M tokens.
Claude Sonnet 5 — explained in more detail
Anthropic released Claude Sonnet 5 on June 30, 2026 as the latest model in the Sonnet line and the successor to Sonnet 4.6. The Sonnet range is Anthropic’s mid-tier: faster and cheaper than the flagship Opus models, but more capable than the small Haiku models. Anthropic positions Sonnet 5 as the most agentic Sonnet so far — a model that makes plans, uses tools such as browsers and terminals, and runs longer autonomous workflows that until recently required much larger models.
The core claim: performance approaches Opus 4.8 while the price stays at Sonnet level. Anthropic reports gains in reasoning, tool use, coding and knowledge work, along with lower hallucination and sycophancy rates than its predecessor.
Example / Practical use
Sonnet 5 is aimed above all at agentic applications — systems that plan multiple steps and drive tools rather than returning single answers. In Anthropic’s evaluations the model improves in particular on agentic search (BrowseComp) and computer use (OSWorld-Verified). This makes it well suited to cost-sensitive agent workflows, where many model calls accumulate and per-token price scales.
The key data: pricing is $2 per 1M input tokens and $10 per 1M output tokens (made permanent on August 10, 2026). Access is proprietary via the Claude API (model ID claude-sonnet-5), Claude.ai (the default model for Free and Pro plans), and via AWS and Microsoft Foundry, with Google Vertex following later. On Claude.ai, Sonnet 5 replaced the previous default model Sonnet 4.6.
Distinction
Sonnet 5 is not the family’s flagship — that role belongs to Opus 4.8 at release. Sonnet is deliberately the balance tier between cost and capability: where maximum capability matters you reach for Opus, where minimal latency and cost matter you reach for Haiku. Sonnet 5, however, narrows the gap to Opus and makes tasks that previously needed a top model accessible at the Sonnet rate.