Claude Sonnet 5.5 ships: 30% faster, up to 30% cheaper per task

Anthropic released Claude Sonnet 5.5 on September 28 as the second model in its Claude 5.5 family, and its launch post makes two concrete claims against the previous Sonnet 5: the new model runs more than 30% faster, and it costs up to 30% less "for most work." The company also calls it a "clear upgrade" over Sonnet 5.

The post does not say how those figures were measured, which tasks count as most work, or whether the saving comes from a lower price or from the model using fewer tokens to finish a job. Its linked page and attached video were not collected for this explanation, so those details are not covered here. As written, the speed and cost numbers are Anthropic's own account.

An independent measurement complicates the cost claim. Artificial Analysis reported the same day that Sonnet 5.5 scored 56 on its Intelligence Index, two points behind Claude Opus 5.5 at max effort, but used more output tokens per task than any model it had measured. By its account, Anthropic kept Sonnet 5.5's list prices the same as Sonnet 5's, and the extra output put Sonnet 5.5's cost at $7.60 per task on that index, about 50% higher than Sonnet 5's. Artificial Analysis measured its own task set, with the record token use at the highest effort setting, and tested a pre-release version with a bug that Anthropic said was fixed for the public release. Anthropic does not say what work its figure covers, so the two accounts do not directly contradict each other, but the up-to-30% saving cannot be assumed for every workload.