Anthropic releases Claude Sonnet 5.5 — faster, half the price and harder to copy

Anthropic has released Claude Sonnet 5.5, a new version of its mid-tier model that edges closer to its flagship performance. According to The Next Web (September 28), the model scores 70.6% on the Terminal-Bench 4.0 agentic coding benchmark, versus 66.4% for the pricier Opus 5.5 and 10.3% for Sonnet 5.

Half the price, harder to copy

Claude Sonnet 5.5 costs about half as much per token as Opus 5.5. It is also the first Sonnet model to ship classifiers that stop its reasoning being extracted, weeks after researchers decoded 315,320 thinking blocks from public agent traces.

Why it matters

The launch sharpens competition in the coding-model segment, where performance-per-dollar is becoming the deciding factor for developers and product teams.

Source: The Next Web — Anthropic releases Claude Sonnet 5.5