Anthropic released Claude Opus 5 on Friday at $5 per million input tokens and $25 per million output tokens, matching Opus 4.8 on price and undercutting the company’s own flagship Fable 5 by half. The pricing decision is the point of the launch, and it’s a decision about which customer the company thinks is actually paying the bills.
“Enterprises, in our feedback and with our customer base, are looking for value,” said Dianne Penn, Anthropic’s head of product management for research. Opus 5 is the fourth model in the Claude 5 family to ship in under two months, following Mythos 5, Fable 5, and Sonnet 5 in June. Only the lightweight Haiku tier is still awaiting its upgrade.
On benchmarks, Opus 5 beats Fable 5 on Frontier-Bench v0.1 and Zapier AutomationBench, and scores three times higher than the next-best model on ARC-AGI 3. It remains behind Mythos 5 on offensive cybersecurity and behind Fable 5 on biology research. That gap is the design.
Anthropic expects its safety classifiers to engage 85 percent less often on Opus 5 than on Fable 5, a direct consequence of the narrower capability profile. A new beta feature, Automatic Fallbacks, routes flagged API requests to a less powerful model instead of refusing them. Fortune reports Opus 5 also ships with a per-task toggle for low, medium, or high effort, letting customers price capability against spend.
The regulatory subtext is unmissable. Fable 5 was pulled on June 12 under an export-control directive and restored on June 30; Anthropic says it continues to work with government agencies on pre-deployment testing. Opus 5 becomes the default on Claude Max with a one-million-token context window, which is another way of saying the model most enterprises will actually touch is the one built to trip fewer wires.