Anthropic says the new model shows strong gains in coding. It can work more reliably across large software projects, review code, debug issues and even detect its own mistakes.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results