The new Claude Sonnet 4.5 model from Anthropic offers considerable enhancements in several areas. The model has improved code generation, code improvement detection, and instruction following. The model has significantly reduced behaviors such as deception. With this latest update, the platform is more resistant to prompt injection attacks. The new model is best for using computers with complex agent building. The new update allows the model to interpret screen content. Members of the premium tier will have access to a new AI model without any price increase.
