Anthropic's Claude Sonnet 5.5 Matches Opus 5.5 on Benchmarks, Costs 30% Less

Claude Sonnet 5.5, released on September 28, 2026, achieves a GDPval-AA score of 1,844, nearly matching Opus 5.5's score of 1,846. Anthropic claims Sonnet 5.5 generates output more than 30% faster than its predecessor, Sonnet 5, which scored only 1,449 on the same benchmark. Additionally, Sonnet 5.5 is reported to cut per-task costs by up to 30%, with input token costs set at $2 per million tokens and output token costs at $10 per million tokens.

In benchmark tests, Sonnet 5.5 scored 70.6% on Terminal-Bench 4.0, a significant improvement from Sonnet 5's 10.3%. Similarly, it scored 55.5% on CursorBench 4.0 compared to Sonnet 5's 34.1%. The FrontierCode 1.1 benchmark showed Sonnet 5.5 at 46.2% versus 42.4% for Sonnet 5, and it achieved a 61.6% score on Chartography, up from 15.6% for Sonnet 5.

Sonnet 5.5 is also noted for its ability to play through Pokémon Red using only screenshots, marking a significant advancement in interactive AI capabilities. Requests involving high-risk cybersecurity tasks are still routed to Sonnet 5, indicating ongoing optimization challenges. The upcoming Claude Haiku 5.5 is expected to build on these advancements in the coming weeks.