The bottom line: Anthropic unveils Claude Opus 4.8, an improved AI model offering better judgment, faster processing, and new features like Dynamic Workflows at the same price as its predecessor. Early testers report significantly higher reliability for agentic tasks.
Anthropic presents Claude Opus 4.8, an improved version of the AI model with greater reliability and better collaboration capabilities. The new model is available immediately and costs the same as its predecessor.
Claude Opus 4.8 builds on Opus 4.7 and demonstrates improvements across all benchmarks. The model proves to be a more effective collaborator for agentic tasks and complex problem-solving.
The update brings several new features. Users on claude.ai can now control how intensively Claude works on a task. Claude Code receives a new “Dynamic Workflows” feature for large-scale problem-solving. The new Fast Mode for Opus 4.8 operates 2.5 times faster and costs three times less than earlier models.
Early testers report noticeable improvements: the model demonstrates better judgment in agentic tasks, asks the right questions, and recognizes its own errors. In benchmark tests, Claude Opus 4.8 outperforms its predecessors and competing models in areas such as coding, reasoning, and practical knowledge tasks. Particularly impressive is its performance on computer use and browser agent tasks, achieving 84 percent on the Online Mind2Web benchmark.
The model’s reliability makes it especially valuable for specialized applications: from legal AI to autonomous engineering work. Users report a noticeable quality leap in speed, context understanding, and consistent execution of complex, multi-step tasks.
Source: www.anthropic.com