Anthropic has released Sonnet 5.5, the latest version of its mid-tier AI model, focusing on faster performance, lower token usage, and more efficient agentic workloads.
The company positions Sonnet 5.5 as a general-purpose work assistant for tasks including coding and creating office documents. It arrives roughly three months after Sonnet 5, which Anthropic had promoted for its ability to run AI agents at lower cost.
Anthropic Says Sonnet 5.5 Is 30% Faster
Speed is the main upgrade in Sonnet 5.5. Anthropic says the model is 30% faster than its predecessor and consumes tokens at a significantly slower rate.
Within Anthropic’s model lineup, Sonnet sits below Opus in overall capability but is designed to be more agile and cost-efficient for certain workloads. According to Anthropic’s benchmarks, Sonnet 5.5 performs better than Opus 5.5 on some agentic coding tasks.
The company attributes that advantage in part to Sonnet 5.5’s ability to run multiple agents without exceeding cost limits, making it potentially useful for coding workflows that involve several concurrent tasks.
Sonnet 5.5 Gets Stronger Cyber Safeguards
Anthropic also says Sonnet 5.5 has cyber capabilities comparable to Opus 5. As a result, the company says this is the first Sonnet model that will be subject to the same cyber safeguards applied to Fable and Opus.
The added protections reflect the model’s increased capability in security-related tasks as Anthropic expands the performance of its mid-tier systems.
New Haiku Model Is Also Coming
Anthropic plans to release an updated version of Haiku, its smallest model, in the coming weeks, although the company has not announced a specific launch date.
Sonnet 5.5 arrives during a period of rapid model releases across the AI industry, as major labs compete on speed, cost, coding performance and increasingly capable agentic systems.
