Anthropic Claude Opus 5 nears Fable at half the price

Opus 5 is less about a single jaw dropping benchmark and more about Anthropic tightening the performance per dollar screws across its lineup. By pushing an Opus tier model closer to Fable 5 on coding and knowledge work while keeping the cheaper Opus pricing, Anthropic is signaling that frontier level capability is migrating down the stack fast. For many enterprise workflows, that is more strategically important than another two points on an abstract reasoning test.

In the race to AGI, this release underscores where we are on the curve: incremental but frequent upgrades, optimized for efficiency, safety and breadth of deployment rather than headline grabbing capability spikes. Anthropic is clearly trying to position Opus as the default “serious work” model for developers who cannot justify Fable pricing, and it is doing so while talking up stronger behavioral alignment and automatic fallbacks when the model hits policy limits. That combination is aimed both at regulators and at customers spooked by OpenAI’s recent agent incident.

The competitive message to OpenAI and Google is simple: you do not just have to beat our best model, you have to beat what we can sell cheaply, safely and often. If Opus 5 delivers on that promise in real workloads, it will pressure rivals to compress their own pricing and refresh cycles, accelerating the march toward frontier like capabilities becoming a commodity.