The competition among AI developers shows no signs of slowing down, and Anthropic has now unveiled the latest iteration of Sonnet, its mid-range offering. According to the company, this new version operates considerably faster—and at a much lower cost—than the model it replaces. Anthropic positions Sonnet 5.5 as a perfect helper for routine work, from writing code to producing office documents.

The previous iteration, Sonnet 5, made its debut roughly three months earlier. Back then, the headline feature was cost-effective agentic deployment—the capacity to operate agents more cheaply than rival offerings. With Sonnet 5.5, the emphasis has shifted squarely to velocity. Anthropic asserts that Sonnet 5.5 runs 30 percent faster than the earlier version and consumes tokens at a notably slower rate.

Within Anthropic's model lineup, Sonnet sits below the more powerful Opus, yet it can sometimes prove more practical thanks to its nimbleness. Notably, Anthropic's benchmarks indicate that Sonnet 5.5 outperforms Opus 5.5 when it comes to agentic coding, probably owing to its capacity to spin up multiple agents while staying within cost boundaries.

Sonnet 5.5 is also credited with substantial cyber capabilities, with the company stating that these are "comparable" to those of Opus 5. Consequently, Anthropic notes that 5.5 marks the first Sonnet model to fall under the same cyber safeguards that govern Fable and Opus.

Additionally, the company intends to roll out a fresh version of Haiku—its smallest model—in the weeks ahead, though no specific launch date was provided.

The past year has brought a wave of new model launches from the leading AI labs. Only last week, OpenAI introduced several new models, among them upgraded editions of Sol and Luna, its mid-tier and budget-conscious options. Meta likewise revealed a new model, which it said would drive an upcoming feature tied to its smart glasses.