Anthropic ships Claude Opus 5, pitching frontier work at half the price
Anthropic launched Claude Opus 5 at the same $5-per-million-input-token and $25-per-million-output-token prices as Opus 4.8, claiming substantially better performance on coding, reasoning and business-automation benchmarks. The model emphasizes self-verification, concise outputs and fewer tool calls, with early customers reporting improvements in software development, legal redlining and due diligence. Anthropic also claims stronger alignment and limits on exploit generation, while keeping Opus 5 behind Mythos 5 for advanced biology and offensive cybersecurity tasks. The article notes that the benchmark results and customer figures are vendor-supplied, urging publishers and newsrooms to test the model against their own workflows. Beta tool-switching and API-fallback features, along with a faster paid mode, accompany the launch.
Reporting styleBalanced
Mostly factual reporting of Anthropic's launch claims, combined with cautious editorial analysis and advice to conduct independent testing.