Anthropic is accelerating its strategy with the upcoming launch of Claude Opus 4.7, a model equipped with a tool capable of generating complete digital products based on simple instructions. This development clearly marks a repositioning towards an automated production platform. At the same time, the presence of more advanced internal models has sparked discussions about the uses and limitations of these systems.
Opus 4.7 will introduce new competition to the creative tools market, allowing users to create websites, presentations, or landing pages using natural language, without the need for advanced technical skills.
This approach reflects a shift towards AI that can replace some functions of traditional specialized software. By centralizing generation, structuring, and deployment, Anthropic aims to significantly simplify the creative process while redefining the standards for no-code and AI-assisted design.
Controversial Capabilities of the Internal Model Mythos
In testing, Mythos succeeded 3 times out of 10 attempts, averaging 22 steps completed out of 32, compared to Opus 4.6's average of 16 steps. These results and on-chain data demonstrate high autonomy in executing complex scenarios, including coordinated actions across multiple phases of attack.
These performances highlight a significant evolution of AI models towards comprehensive operational capabilities, surpassing mere language processing. Mythos showcases the system's shift towards autonomous action in sensitive technological environments.
Meanwhile, this progress also exposes the limitations of current evaluation tools. Several industry participants, including OpenAI, have questioned the reliability of certain benchmarks, arguing that these benchmarks are contaminated or lack representativeness. Test results indicate significant disparities in model performance, with Gemini at 0.37%, GPT-5.4 at 0.26%, while humans achieved 100%.