Anthropic Releases Claude Sonnet 5.5 With 30% Speed Boost and Lower Token Costs
Anthropic says its September 28 release is 30% faster than Sonnet 5 and uses tokens more slowly, while outperforming Opus 5.5 on agentic coding tasks. It also matches Opus 5’s cyber capabilities and receives the same safeguards as Fable and Opus.
Anthropic released Claude Sonnet 5.5 on September 28, 2026, the latest version of its mid-tier AI model, promising faster response times and reduced token consumption compared to its predecessor.
The company describes Sonnet 5.5 as suited for everyday tasks including coding and creating office documents. Anthropic claims the model is 30 percent faster than Sonnet 5 — which launched roughly three months earlier — and burns tokens at a significantly slower rate.
Within Anthropic’s model lineup, Sonnet sits below the more powerful Opus tier but can outperform it in specific use cases. According to Anthropic’s own benchmarks, Sonnet 5.5 performs better than Opus 5.5 on agentic coding tasks, which the company attributes to its ability to spawn multiple agents without exceeding cost limits.
Anthropic also highlighted the model’s cyber capabilities, stating they are comparable to those of Opus 5. Because of this, Sonnet 5.5 becomes the first Sonnet model subject to the same cyber safeguards that apply to Anthropic’s Fable and Opus models.
The company also indicated it plans to release a new version of Haiku, its smallest model, in the coming weeks, though no firm date was provided.
The release comes amid a wave of new model launches across the AI industry. OpenAI released several new models the prior week, including updated versions of Sol and Luna, its mid-tier and budget models. Meta also announced a new model intended to power an upcoming feature for its smart glasses.
For developers and businesses running AI agents at scale, the combination of faster output and lower token costs in Sonnet 5.5 could reduce operational expenses — though real-world performance will depend on individual workloads.