Anthropic Releases Claude Opus 5, Halving Token Costs While Surpassing Fable 5 on Key Benchmarks

The price-performance jump echoes the 1982 launch of the Commodore 64, which undercut rivals while matching their capabilities and forced an industry-wide repricing; today’s AI labs are repeating that cycle in software form. Opus 5’s debut reinforces two long-term currents: rapid commoditisation of frontier intelligence and the growing strategic linkage between model cost, safety policy, and national security export controls—much like the 1990s crypto-wars when stronger but tightly regulated RSA implementations coexisted with ‘export-grade’ ciphers. Over a century horizon, the real significance is not this single model but the acceleration cadence: if near-frontier capability can be halved in cost every few months, diffusion—both beneficial and malicious—becomes inevitable, potentially mirroring how mass-produced rifles erased the advantage of elite armies by 1914. Whether Anthropic’s tightened yet lighter guardrails can scale as fast as capability/cost curves remains the unresolved question that will shape AI governance far more than today’s benchmark leaderboard.