Anthropic ships Sonnet 5.5, says costs drop up to 30%
Claude Sonnet 5.5 runs 30% faster than its predecessor and nearly matches Opus 5.5 on benchmarks, Anthropic says, though the claims are still unconfirmed independently.
Anthropic released Claude Sonnet 5.5 on Monday, the second model in its Claude 5.5 family. The company said it runs more than 30% faster than Sonnet 5 while costing up to 30% less per task. The model is live now under the ID claude-sonnet-5-5, Anthropic said.
https://x.com/claudeai/status/2104633115620823187
The Decoder reported that Sonnet 5.5 nearly matches Opus 5.5 on several benchmarks while undercutting it sharply on price. The outlet noted that independent testing has not yet confirmed Anthropic's own figures. TestingCatalog said the rollout reached Claude's apps and API within the hour of the announcement.
https://x.com/testingcatalog/status/2104632421450764779
Anthropic put Sonnet 5.5's score on the Terminal-Bench 4.0 coding benchmark at 70.6%, against 10.3% for Sonnet 5. Its GDPval-AA knowledge-work score reached 1,844, within two points of Opus 5.5's 1,846. Per-token pricing stays at $2 per million input tokens and $10 per million output tokens, unchanged from Sonnet 5. The saving comes from using fewer tokens per task, the company said.
The model reached Amazon Web Services, Google Cloud and Microsoft Azure alongside Anthropic's own apps and API on release day, the company said. Anthropic described Sonnet 5.5 as a faster, cheaper complement to Opus 5.5. It suits well-scoped tasks such as fixing bugs and producing documents, slides and spreadsheets, the company said.
The commentator kimmonismus noted that Sonnet 5.5 lands at the same per-token price as OpenAI's GPT-6 Sol. That sets up a direct comparison between the two companies' mid-tier models once independent benchmarks appear.