Claude Opus 4.7 Reportedly Achieves 92% Honesty Rate in Hallucination Benchmarks

Summary

The Economic Times reports that Anthropic's Claude Opus 4.7 has achieved a 92% honesty rate, a metric tied to reduced hallucination in AI-generated outputs. The article frames this as a significant step toward human-like AI reliability, though the specific benchmark methodology and evaluation framework referenced in the reporting are not fully detailed in the available content. This development is directly relevant to Anthropic's competitive positioning in the foundation model market, particula