Anthropic ships Claude Sonnet 5.5: what happened and why it matters

What happened: Anthropic released Claude Sonnet 5.5 on 28 September 2026, a week after Claude Opus 5.5. It is the second model in the Claude 5.5 family, built as a faster, cheaper companion to the flagship rather than a replacement for it.

The price: Pricing is unchanged from Sonnet 5, at $2 per million input tokens, $10 per million output tokens and 20 cents per million cache reads. Anthropic did not touch the rate card at all for this release.

The speed and cost: Outputs run over 30% faster than Sonnet 5, making it Anthropic's fastest Sonnet model yet, and typical tasks cost up to 30% less overall, from fewer tokens and tool calls rather than any rate cut.

The benchmarks: Sonnet 5.5 scored 1844 on Anthropic's GDPval-AA v2.1 test, just behind Opus 5.5's 1846 and far above Sonnet 5's 1449. It also hit 80.1% on OSWorld 2.1 and 64.5% with tools on Humanity's Last Exam.

Best for: Anthropic says Sonnet 5.5 is strongest at well-scoped everyday tasks — fixing bugs, and producing documents, slides and spreadsheets — leaving deeper, harder reasoning work to Opus 5.5 instead.

Where it runs: The model is available right now on Amazon Web Services, Google Cloud and Microsoft Azure as claude-sonnet-5-5, and is already live inside Claude Code, Anthropic's coding tool.

Safety detail: It is the first Sonnet model ever built with the cyber safeguards Anthropic uses on its most capable models, with a cyber capability it says is comparable to Opus 5's. Its output also carries an invisible watermark, added for rules including the EU AI Act, SiliconANGLE reported.

Why it matters: Anthropic is closing the gap between its mid-tier and flagship models while holding prices completely flat, giving buyers near-flagship scores on its own benchmarks at mid-tier prices.

What's next: Anthropic's Opus 5.5 announcement said Haiku 5.5, the cheapest tier in the family, would follow within weeks; it has not yet shipped.

The line-up: Sonnet sits between the Opus flagship and the smaller Haiku line, so this release moves the middle of Anthropic's range onto the 5.5 generation, six days after the top of it moved.

The takeaway: two launches in six days show Anthropic moving quickly to refresh its whole model line-up at once, rather than staggering upgrades across a year.

The gap left: with Haiku 5.5 still unreleased, one more model is needed before the Claude 5.5 family Anthropic promised is complete. Anthropic also says Sonnet 5.5 writes more clearly and more like a person than its predecessor.

Bottom line: Sonnet 5.5 costs the same as Sonnet 5 and, on Anthropic's figures, runs faster and uses fewer tokens per task, while Opus 5.5 stays the pick for the hardest work.