Anthropic Launches Cloud Opus 4.5
Artificial intelligence (AI) technology is advancing rapidly these days. As a result, AI companies around the world are launching new and powerful AI models. In this series, Anthropic has now launched Cloud Opus 4.5, which the company describes as the most advanced and powerful model in its Cloud 4.5 family.
This is the same model that Anthropic will now use as its top commercial AI model. Interestingly, the company has not only increased performance in this version but also significantly reduced token usage, reducing both cost and compute.
Anthropic has clearly stated that Opus 4.5 brings upgrades in three major areas: code generation, agentic tools, and long-context reasoning. All three areas feature significant improvements over the previous model, Cloud Opus 4.1.
A Big Leap in Coding
Opus 4.5 can now complete complex and lengthy coding tasks in fewer tokens than before. Anthropic claims that the model produces the same output in approximately 65 percent fewer tokens than before. This means the model has become more efficient at writing, understanding, and refactoring code. The company attributes this improvement to improved planning and internal reasoning.
A New Level of Multi-Step Workflows
The new Opus 4.5 is also capable of handling tasks that require handling multiple processes simultaneously. In internal tests, it was able to refactor two different codebases simultaneously and coordinate three agents.
This model creates high-level plans and, at the same time, delves into the details and completes low-level tasks. The company says that thanks to the new system, it eliminates unnecessary tool loading and only calls those necessary for the actual task, reducing context usage by approximately 85 percent.
Large Files and Creative Content
Opus 4.5 handles longer documents and narrative content better. Anthropic claims that the model maintains character, tone, and plot continuity in stories of up to 10 to 15 pages without disrupting them. Furthermore, it is more capable of 3D reasoning, understanding and interpreting scenes and spatial relationships in greater detail.
