Anthropic releases Claude Sonnet 5.5: second Claude 5.5 model; faster, lower-cost complement to Opus 5.5

Why it matters

【Official release】

Anthropic released Claude Sonnet 5.5 on 2026-09-28 — the second model in the Claude 5.5 family after Opus 5.5. Canonical page: https://www.anthropic.com/claude-sonnet-5-5 (also featured on the Newsroom). Page eyebrow is day-level (September 28, 2026); CMS siteSettings._updatedAt≈2026-09-28T17:58:34Z is a verifiable approximate go-live; cross-checked with @claudeai post at 2026-09-28T18:03:31Z. API id: `claude-sonnet-5-5`; Anthropic says it is available on the Claude Platform and on Amazon Web Services, Google Cloud, and Microsoft Azure.

【Verifiable points】

Official claims vs Sonnet 5: 30%+ faster output generation; up to ~30% lower cost for most work at the same list price ($2/$10 per MTok input/output; cache reads $0.20) via fewer tokens on typical tasks. Positioned as a faster, lower-cost complement to Opus 5.5 for well-scoped everyday work, bugfix, and polished docs/slides. Vendor-reported Terminal-Bench 4.0 (agentic coding) 70.6% vs Sonnet 5’s 10.3%; GDPval-AA about two points below Opus 5.5. First Sonnet to launch with cyber safeguards comparable to Opus 5 (cyber tasks visibly fall back to Sonnet 5). Haiku 5.5 is still described as forthcoming for high-volume/cost-sensitive use.

【Our take】

Site already has e/666 (Opus 5.5 official) and e/682 (Opus/Astra opinion); no Sonnet 5.5 event found → new official item. Mid-tier Claude 5.5 family release with material impact on everyday/agent toolchain cost-speed structure → surfaces include home_week. Facts limited to official visible text; benchmarks are vendor-reported and not restated as cross-lab comparisons. Do not mix in DeepSeek V5/V4.1 rumor threads.