Claude Sonnet 5.5 is 30% faster and competes on task cost, not token price

Anthropic has released Claude Sonnet 5.5 as a faster, more economical counterpart to Opus 5.5. Its price list is unchanged from Sonnet 5, but the company says the model completes most work with fewer tokens and generates output over 30% faster.

The same token prices conceal a lower cost per completed task

Sonnet 5.5 costs $2 per million input tokens, $10 per million output tokens, and $0.20 per million cache-read tokens. In Anthropic's own tests, it cost up to 30% less per task than Sonnet 5.

The model scored 70.6% on Terminal-Bench 4.0, compared with 10.3% for its predecessor. At Max effort it scored 46.2% on FrontierCode 1.1 and 55.5% on CursorBench 4.0. It is available on the Claude Platform as claude-sonnet-5-5 and through AWS, Google Cloud, and Microsoft Azure.

Agent efficiency is measured in finished work, not cheap tokens

For coding agents, fewer tool calls and a shorter path can outweigh list-price differences. A model with the same token price is cheaper when it repeats fewer searches, wanders less through a repository, and completes more tasks without human intervention.

Anthropic positions Sonnet for routine, well-scoped work while reserving Opus 5.5 for open-ended tasks requiring sustained judgment. That is a routing guide for product teams, not an automatic reason to replace one model everywhere.

The dramatic benchmark jump still comes from the vendor

The figures compare different effort settings, and some evals used a pre-release deployment. Anthropic itself says Opus remains stronger on complex, open-ended work. Because Sonnet 5.5 has stronger cyber capabilities, it also uses new safeguards and may fall back to Sonnet 5 for higher-risk tasks.

Internal evals will show whether 30% survives production

Teams should compare full task cost, time to an accepted change, human interventions, and failed tool calls. Only the same workflow on their own data will show whether faster generation produces faster delivery.

Lilith's verdict

Sonnet 5.5 did not replace the token price tag; it shortened the walk to the checkout. If company evals confirm fewer steps and human interventions, accountants will notice before benchmark fans do.