Benchmark Leader Change: Anthropic Rolls Out Claude Opus 5 Model at Old Price

The race of frontier models continues at a relentless pace. On August 7, 2026, the Anthropic lab officially introduced `Claude Opus 5`—its most powerful release designed for complex coding and deep analytical work.

According to the stated technical specifications, Opus 5 sets new absolute records on the industrial Frontier-Bench and GDPval-AA tests, confidently outperforming not only its predecessor (Opus 4.8) but also the latest releases from competitors. The main news for the B2B market is the aggressive pricing policy. Anthropic is launching a SOTA (State-of-the-Art) model across all platforms without raising the cost of API calls. This is a retaliatory strike against OpenAI's recent dumping. Corporate clients get a radical boost in intelligence within their existing OPEX budgets, which further squeezes the margins of the AI labs themselves amid ultra-expensive inference.

Source: Anthropic