Anthropic Launches Claude Sonnet 5 as a Lower-Cost Model for AI Agents

New Claude Model Targets Agentic AI Workflows

Anthropic has launched Claude Sonnet 5, a new version of its midsize AI model designed to run AI agents at a lower cost.

The release comes as major AI companies compete to make their models better at handling longer, more complex tasks with less human direction. Claude Sonnet 5 is built to plan, use tools, work across multiple steps, and complete tasks more independently than earlier Sonnet models.

Sonnet 5 Offers Stronger Performance at a Lower Price

Anthropic is positioning Claude Sonnet 5 as a more affordable option for developers and businesses that want to build or run AI agents. The company says the model delivers performance close to its more powerful Opus model while costing significantly less.

At launch, Claude Sonnet 5 is priced at $2 per million input tokens and $10 per million output tokens through August 31. After that period, the price is set to rise to $3 per million input tokens and $15 per million output tokens.

The model is now the default option for Claude Free and Pro users and is available across all Claude subscription plans.

Built for Coding, Tool Use, and Knowledge Work

Claude Sonnet 5 shows improvements in several areas important for AI agents, including reasoning, coding, software tasks, tool use, and knowledge work.

Anthropic says the model performs better than Claude Sonnet 4.6 on agent-focused tasks. It is also designed to complete more complex jobs that may have caused earlier models to stop before finishing.

This makes Sonnet 5 especially relevant for developers building agents that need to browse, use terminals, update systems, generate code, check their own work, or complete business workflows from start to finish.

AI Companies Push Toward Cheaper Agent Models

The launch reflects a larger shift in the AI industry. Agentic capabilities are no longer being limited to the most expensive frontier models. Instead, companies are working to bring planning, tool use, and autonomous task completion into cheaper and faster models.

Anthropic’s move follows similar efforts from rivals such as OpenAI and Google, which have also promoted newer models as more capable of handling agent-style work.

As competition grows, cost is becoming a major factor. Developers and companies want models that can complete complex tasks reliably without making agent workflows too expensive to run at scale.

Safety Remains a Major Focus

Anthropic says Claude Sonnet 5 also improves on safety compared with its predecessor. The company says the model is better at refusing harmful requests, resisting misuse, and handling prompt-injection attacks.

The model is not positioned as Anthropic’s most powerful or safest system overall, but it is designed to offer a stronger balance between capability, cost, and reliability for everyday agent use.

What This Means for Developers

For developers, Claude Sonnet 5 could make it cheaper to build AI agents that perform practical tasks across coding, operations, customer workflows, and business automation.

The model gives users another option between low-cost lightweight models and more expensive high-end models. That balance may make Sonnet 5 useful for teams that need strong performance but also want to control infrastructure and API costs.

With Claude Sonnet 5, Anthropic is signaling that the next phase of AI competition will not only be about which model is smartest. It will also be about which model can do useful work at the best price.