Anthropic Releases Claude Sonnet 5.5: 70.6% on Terminal-Bench 4.0 at the Same $2/$10 Price

Anthropic has introduced Claude Sonnet 5.5, the second model in its Claude 5.5 family. The release targets developers needing high performance at a lower cost tier, positioning the model between the flagship Opus 5.5 and previous Sonnet iterations.

The model scores 70.6% on Terminal-Bench 4.0 and lands within two points of Opus 5.5 on GDPval-AA, suggesting competitive agentic and coding capabilities. Anthropic also reports that Sonnet 5.5 produces output more than 30% faster than Sonnet 5, addressing latency constraints for production applications.

Pricing remains unchanged at $2 per million input tokens and $10 per million output tokens, reinforcing Anthropic's strategy to compete on price-to-performance rather than undercutting rivals. This makes the model attractive for high-volume inference workloads where cost efficiency is critical.

The release tightens the gap between Anthropic's mid-tier and flagship models, potentially pressuring competitors like OpenAI and Google to justify premium pricing for marginal gains. It also signals that benchmark parity with top-tier models is increasingly achievable at lower price points.

For developers building agents or coding assistants, Sonnet 5.5 offers a practical middle ground: near-flagship quality with faster generation and lower cost. The unchanged pricing structure may accelerate adoption in enterprise pipelines where token economics drive deployment decisions.

This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.