Claude Sonnet Glossary

The mid-tier of Anthropic's *Claude* family — balanced capability and cost. The default model for most agentic-tool-use workflows.

Claude Sonnet is the middle tier of Anthropic’s Claude model family, sitting between the most-capable Opus and the fast-and-cheap Haiku. Sonnet was first introduced with Claude 3 in March 2024, and Sonnet 3.5 (June 2024) became the breakout model that established the tier as the practical default for most production agentic workloads — including, notably, the *Claude Code * product’s underlying engine through much of 2024–2025. Current versions include Sonnet 4, 4.5, 4.6, and 4.7.

For this Dictionary, Sonnet is the practical workhorse. Where Opus is reserved for tasks where the marginal capability gain materially matters (long-form careful writing, complex multi-step reasoning, the entries you are reading right now), Sonnet covers most everything else: tool use, sub-agent dispatch, routine drafting, summarisation, code edits at moderate complexity. The Dictionary’s cost-of-operation argument in Opus Addict treats Sonnet as the escape valve — the tier most operators should default to and reserve Opus for the marginal hard cases.

Sonnet 3.5 is also the model on which *Claude Code * reached its first signs of product-market fit per Boris Cherny’s *Coding Solved * talk, before the May 2025 Opus 4 release inflected the exponential-growth curve the product is currently on.

See also

Return to Dictionary All Entries (A–Z) For Students Other Writing