Claude Sonnet 5.5 targets speed and task cost
Anthropic has introduced Claude Sonnet 5.5, the second model in its Claude 5.5 family. The company says it generates outputs more than 30% faster than Sonnet 5 and can cost up to 30% less per task for most work. Token prices are unchanged: $2 per million input tokens, $10 per million output tokens, and $0.20 per million cache-read tokens. The claimed task savings come from using fewer tokens to do comparable work.
Anthropic positions the model for well-scoped, everyday work: fixing bugs and producing documents, slides, and spreadsheets. In the company’s published results, its Terminal-Bench 4.0 score rises to 70.6%, compared with 10.3% for Sonnet 5. On the GDPval-AA knowledge-work evaluation, its result is close to Opus 5.5. Anthropic also reports gains in image understanding, long-horizon tasks, and computer use.
Benchmark gains do not erase the judgment gap
These figures come from Anthropic’s evaluations and tests; they cannot establish how the model will perform across all real-world work. The company explicitly says Opus 5.5 remains clearly stronger on complex, open-ended tasks requiring sustained judgment. Sonnet 5.5 is framed as a faster, lower-cost complement, not a universal replacement. Feedback from early testers is useful context, but it is not a guarantee of results for every user or workflow.
On safety, Anthropic says Sonnet 5.5 receives cyber safeguards and fallbacks similar to those used for its most capable models because its cybersecurity abilities are comparable to Opus 5. Its biology safeguards are unchanged from Sonnet 5. The company says these protections target a narrow set of high-risk requests and do not affect routine software development or most life-sciences work. For adopters, quality, latency, and actual task costs still need to be checked in the workflows they intend to use.