Sonnet 5.5 keeps token prices while cutting usage per task

Anthropic attributes savings to fewer tokens rather than a lower base rate.

Capvise MediaEvent 28 Sept 2026Published 15 Oct 20261 min read

Photo: Context photograph of real server racks: NOIRLab HQ Server Racks (6V6A0395-CC). The facility is not identified as belonging to the company or product in this report. Credit: NOIRLab/NSF/AURA/T. Slovinský, “NOIRLab HQ Server Racks (6V6A0395-CC).jpg,” Wikimedia Commons, CC BY 4.0. Adapted by resizing and WebP re-encoding. · CC BY 4.0 · Resized for web delivery

Same rate, different consumption

Anthropic introduced Claude Sonnet 5.5 on 28 September, keeping prices at $2 per million input tokens, $10 per million output tokens and $0.20 for cache reads. The company reported output generation more than 30% faster and task costs up to 30% lower than Sonnet 5 in its tests.

The claimed saving comes from doing the work with fewer tokens, rather than changing the listed input and output rates. Anthropic positioned Sonnet for well-scoped tasks and iteration, while retaining Opus 5.5 for more open-ended work requiring sustained judgement. It also added stronger cyber safeguards. For application teams, the useful comparison includes task completion, token consumption and latency on their own workloads; equal token prices do not imply equal cost per completed request.