Why You Should Stop Trusting Flawed Code: Anthropic Launches Sharper Claude Opus 4.8
SWE-Bench Pro score hits 69.2%, coding errors pass unremarked four times less often, and execution runs 2.5x faster in fast mode. Anthropic’s newest flagship directly targets developer frustration....