Anthropic launches Claude Sonnet 5.5, the AI that runs faster and costs less

Anthropic is not letting up on AI announcements. Less than a week after Claude Opus 5.5, the American startup launched Claude Sonnet 5.5, an artificial intelligence that stands out for its speed.According to Anthropic, Sonnet 5.5 generates its responses more than 30% faster than Sonnet 5.

Claude Sonnet 5.5 is now available: https://t.co/DuxGkiPLRY

— Anthropic (@AnthropicAI) September 28, 2026

Furthermore, the model also costs up to 30% less for most tasks. Yet the rates listed by Anthropic are not changing. The savings come from how the AI operates. It requires far fewer usage tokens to complete the same task.

Also read: Claude may have discovered a DNA secret, and no one knows what it is for

The cost-effective complement to Claude Opus 5.5

Anthropic presents Sonnet 5.5 as the cost-effective complement to Opus 5.5. The model is primarily designed to meet users’ everyday needs. It excels at bug fixing, document generation, and creating presentations or spreadsheets. For its part, Opus 5.5 remains the benchmark for complex work requiring advanced reasoning capabilities.

On paper, performance figures are up, according to the various benchmarks published by Anthropic. Several partner companies have already tested the model ahead of its release, with a marked improvement in efficiency and speed, at lower cost. In terms of security, Sonnet 5.5 has cybersecurity capabilities comparable to those of Claude Opus 5.That’s why it is the first model in the Sonnet range to come with safeguards similar to those of Opus 5.5.

Strong safeguards

In practice, requests deemed riskier are automatically redirected to Sonnet 5. It is nevertheless still possible to ask the AI to fix bugs found in code. The model also includes filters against “distillation,” a technique that allows hackers to copy a model’s capabilities. Anthropic says that three Chinese AIs have already engaged in this practice. It is a common and legitimate technique that laboratories use to create lighter and less expensive versions of their own AI models. It simply becomes problematic when a competitor uses it without any authorization to copy the capabilities of a rival model.

Also read: You thought you were talking to DeepSeek and Kimi, but it might have been Claude – Anthropic’s revelations

Claude Sonnet 5.5 is available now on Anthropic’s platforms, as well as on Amazon Web Services, Google Cloud and Microsoft Azure. Anthropic says it has no intention of stopping there. A Haiku 5.5 version, designed for high-volume use cases, is expected to follow “in the coming weeks.”.

Despite the warnings issued by its CEO Dario Amodei and the calls for a slowdown that are growing throughout Silicon Valley, Anthropic continues to roll out new high-performing models. The startup is likely seeking to reassure the markets ahead of its IPO, which should take place sometime in October.

In its IPO prospectus, Anthropic presents AI as a technology capable of transforming the global economy more profoundly than industrialization, electricity, or the Internet. The company says it wants to build powerful systems that can be controlled. However, it warns that this AI race is extremely costly, particularly due to the development of models, chips, servers, and cloud contracts, which are driving up its spending. Paradoxically, Anthropic does not hesitate to stress once again the risks associated with more autonomous AIs.

👉🏻 Follow the latest tech news in real time: add 01net to your sources on Google, and subscribe to our WhatsApp channel.