Anthropic Releases Claude Sonnet 5.5 as AI Competition Intensifies

SAN FRANCISCO, Sept 29: Artificial intelligence company Anthropic has expanded its Claude 5.5 series with the release of Claude Sonnet 5.5, introducing a faster and lower cost model aimed at businesses and everyday professional workloads.

The company announced the model on September 28, making it the second addition to its Claude 5.5 family after Claude Opus 5.5 was launched the previous week. Anthropic said Sonnet 5.5 is designed to handle tasks such as software development, document creation, spreadsheet work and other business applications.

The launch comes as AI companies increasingly compete to provide models that combine advanced capabilities with lower operating costs. While the most powerful systems are designed for complex reasoning and specialised workloads, businesses are also seeking models that can perform routine tasks quickly without requiring the highest level of computing resources.

Anthropic said Sonnet 5.5 runs more than 30% faster than its predecessor and costs up to 30% less for most work. The company described it as a complement to Opus 5.5, with Sonnet positioned for more broadly defined professional tasks.

The pricing structure remains unchanged from Sonnet 5, at $2 per million input tokens and $10 per million output tokens, according to Reuters. Anthropic said the newer system can complete the same work using fewer tokens, which can affect the overall cost of operating applications built around the model.

For businesses, the ability to reduce the amount of computing required for individual tasks is becoming increasingly important. Companies using AI at scale may process millions of requests, meaning even modest efficiency improvements can have an impact on operating expenses.

Anthropic has positioned the new model particularly toward enterprise users. The company said enterprise customers account for about 80% of its business, with Salesforce, Databricks, Goldman Sachs and Novo Nordisk among its customers.

The company’s focus on enterprise applications reflects the broader transformation of workplace software. AI systems are increasingly being incorporated into coding platforms, office applications, research tools and corporate workflows.

Software development is one of the areas where Sonnet 5.5 is expected to have a significant role. AI coding systems can help developers identify problems, generate code, explain existing programs and work through multiple stages of software projects.

Anthropic said Sonnet 5.5 has improved performance on agentic coding evaluations. Its own testing reports that the model scored 70.6% on Terminal-Bench 4.0, compared with 10.3% for Sonnet 5.

The model is also designed to support longer tasks. Instead of responding to isolated questions, newer AI systems can work through a sequence of related instructions and maintain context while completing a broader assignment.

This capability is becoming increasingly relevant to companies experimenting with AI agents. Such systems can interact with software tools, analyse information and complete multiple stages of a workflow.

At the same time, greater autonomy has increased attention on AI safety. Anthropic’s latest product launch comes during a period of heightened discussion about the risks associated with increasingly capable models.

Chief Executive Officer Dario Amodei earlier called for the AI industry to slow the pace at which new capabilities are released, citing concerns about safety. The company has continued developing new systems while conducting additional testing and introducing safeguards.

The release of Sonnet 5.5 therefore illustrates the tension facing AI developers. Companies are attempting to make their models more capable and useful while also addressing concerns about how those systems could behave when given access to tools or sensitive information.

Anthropic’s approach has been to develop different models for different levels of complexity. Opus 5.5 is aimed at demanding work requiring greater reasoning capabilities, while Sonnet 5.5 is intended to provide a balance between performance, speed and cost.

The company is also preparing another addition to the series. Anthropic said Claude Haiku 5.5, designed for fast and high-volume workloads, is expected to arrive soon.

The three-model strategy could allow businesses to select systems based on the nature of individual tasks. More complicated assignments can be directed toward higher-capability models, while routine requests can be handled by faster and less expensive systems.

This approach could also influence how companies manage their AI budgets. Instead of using one model for every application, organisations can allocate different workloads according to performance requirements.

The development of smaller or more efficient models is also important because AI adoption is expanding beyond large technology companies. Businesses in finance, healthcare, retail, manufacturing and professional services are experimenting with AI tools for internal operations and customer-facing products.

For these organisations, predictable costs and response times can be just as important as benchmark performance.

Anthropic’s latest release comes as the company prepares for a potential initial public offering. Reuters reported that the AI lab is expanding its product lineup as it moves toward a planned IPO.

The company’s growing enterprise business is likely to be an important part of its expansion strategy. Enterprise contracts can provide recurring revenue while giving AI developers access to large-scale use cases that demonstrate how their systems perform outside controlled testing environments.

However, the competitive landscape remains crowded. Anthropic is competing with companies including OpenAI, Google and other developers that are releasing increasingly capable models and integrating them into cloud and productivity platforms.

The rapid pace of new releases means model performance is changing quickly. Features that were considered advanced only a short time ago are increasingly becoming standard expectations for business users.

Anthropic’s decision to keep Sonnet 5.5 at the previous price while improving speed and efficiency also reflects pressure to deliver greater value to customers.

The company said the model is already positioned as a practical option for everyday professional work. Its capabilities include coding, document preparation, spreadsheets and other tasks where users need a combination of accuracy, speed and flexibility.

The next phase of competition may therefore focus less on simply producing larger models and more on delivering systems that can perform useful work efficiently.

With Sonnet 5.5, Anthropic is seeking to strengthen its position in that market by offering a model designed to handle substantial professional workloads without the cost structure associated with its most powerful systems.

The company plans to continue expanding the Claude 5.5 family with Haiku 5.5, giving customers additional options as demand for AI-powered business applications grows.