Anthropic has launched Claude Sonnet 5.5, the second iteration of its Claude 5.5 family. This new model generates output more than 30 percent faster and costs up to 30 percent less per task while nearly matching Opus 5.5 on knowledge-work benchmarks. Performance on Terminal-Bench, a coding assessment, jumps from 10.3 to 70.6 percent. With Haiku 5.5 arriving in coming weeks, Anthropic will soon possess a direct counterpart to each of OpenAI’s three GPT-6 models.
The shift matters because it alters the economics of high-end reasoning without requiring users to switch to the most expensive tier. Companies previously forced to pay premium rates for complex tasks can now access comparable quality at a lower price point. This compression of the performance gap between tiers reduces the financial risk associated with deploying large language models in production environments.
* Terminal-Bench scores rise from 10.3 to 70.6 percent
* Output generation speeds increase by over 30 percent
* Costs drop by up to 30 percent per task
