Anthropic Launches Claude Sonnet 5.5 With Same Token Prices

Anthropic introduced Claude Sonnet 5.5 on September 28, 2026, describing it as a clear upgrade over Sonnet 5 that runs more than 30 percent faster and costs up to 30 percent less for most work. The company frames it as a faster, lower-cost complement to Claude Opus 5.5, aimed at well-scoped everyday tasks, bug fixes and polished documents, slides and spreadsheets, with a Haiku 5.5 model still to come. Sonnet 5.5 is not meant to push the capability frontier the way Opus 5.5 does; Anthropic’s own testers still rate Opus 5.5 stronger on complex, open-ended judgment calls.

On coding benchmarks the jump is large. Anthropic reports Sonnet 5.5 scoring 70.6 percent on Terminal-Bench 4.0, a command-line agentic coding test, against 10.3 percent for Sonnet 5, and within two points of Opus 5.5 on GDPval-AA, a benchmark spanning dozens of occupations. Several partner quotes cited by Anthropic, including from Slack, Zendesk and Box, describe fewer output tokens and faster ticket resolution when swapping in the new model without changing prompts.

The pricing page keeps Sonnet 5.5 at the same per-token rate as Sonnet 5: $2 per million input tokens, $10 per million output tokens and $0.20 per million tokens for cache reads, alongside a $2.50 five-minute cache write rate. Anthropic’s argument is that the model simply needs fewer tokens to finish comparable work, so the effective cost per finished task falls even though the sticker rate does not move. That mirrors a pattern this site’s own price table has already recorded for the Sonnet line, where a scheduled rate increase for the prior standard price did not go ahead.

On safety, Anthropic says Sonnet 5.5 is the first Sonnet model to launch with cyber safeguards and fallbacks resembling those built for its most capable models, because its cybersecurity capability has grown close to Opus 5’s. Its biology safeguards stay unchanged from Sonnet 5, and the company says routine software development and most life sciences work are unaffected by either set of restrictions.

The numbers

• 30%+ faster output generation than Sonnet 5, Anthropic’s fastest Sonnet model to date.
• 70.6% on Terminal-Bench 4.0, versus 10.3% for Sonnet 5.
• $2 per million input tokens and $0.20 per million cache-read tokens, both unchanged from Sonnet 5’s published rate.

For a team already budgeting around Sonnet’s published price, the practical decision this launch creates is whether to keep prompts and workflows as they are and simply bank the token savings, or to redesign agentic pipelines around a model that Anthropic says batches tool calls more efficiently and needs fewer steps to reach the same result.