MODEL ECONOMICS · JULY 24 RELEASE
On July 24, Anthropic shipped Claude Opus 5 at $5 / $25 per million tokens, exactly half of Fable 5's $10 / $50, while landing within 0.5% of it on agentic coding and beating it outright on computer-use automation. This is a small working calculator using Anthropic's own published pricing and benchmark deltas, not marketing language, to see what that spread is worth on an actual monthly workload.
Same monthly workload, two models
Pick a workload and set what you currently spend on Fable 5 in a month. The multiplier applied is Anthropic's own published cost delta for that benchmark, not a flat half-price assumption.
FABLE 5
$1,200
$10 / $50 per Mtok
OPUS 5
$600
$5 / $25 per Mtok
SAVED
$600
50% less per month
CursorBench 3.2: Opus 5 lands within 0.5% of Fable 5's score on this benchmark. Cost multiplier (0.5x) and pricing ($5/$25 Opus 5, $10/$50 Fable 5, Fast mode at 2x base price and ~2.5x speed) are as published in Anthropic's July 24, 2026 Opus 5 announcement. This is a spend estimate against a number you set yourself, not a live billing tool.
WHY THIS MATTERS FOR CLIENT WORK
Nearly every hour I bill runs through Claude Code, so a model price is not an abstract benchmark line, it is a cost lever on every retainer. The useful number here is not that Opus 5 is cheaper, it is that it is cheaper at close to parity on the exact workload I run most, agentic coding, and it is cheaper by more, a third of the cost, on computer-use automation, the category client work is moving toward. That is the argument for switching defaults, not a press release line about being close to the frontier.
Built 27 July 2026 · pricing and benchmark deltas sourced from Anthropic's Opus 5 announcement