Rork adds Claude Opus 5 for coding and visual output
By Steven Van ·
Rork's AI app builder gained Opus 5 on launch day, citing gains on coding benchmarks and first-pass UI quality.
Rork, the tool that turns a description into a production-ready mobile app, added Claude Opus 5 on launch day, and flagged the two things that matter most for its use case: it's state of the art on coding benchmarks, and it has stronger visual output than any previous Opus.
Coding and visuals, the two axes that matter for apps
Building a mobile app is a dual challenge: the code has to be correct (navigation, state, data) and the app has to look right (layout, components, polish). Rork calling out both is telling. On code, Opus 5 roughly doubled Opus 4.8's Frontier-Bench score (43.3% vs 18.7%), the benchmark for real multi-file engineering. On visuals, Rork's own claim of stronger output than any previous Opus points to better first-pass UI, which for an app builder is the difference between a prototype you'd demo and one you'd bin.
Frontier intelligence at half the price
Opus 5's headline is roughly Claude Fable 5-level intelligence at half the price ($5/$25 per million tokens), with a low/medium/high effort toggle. Rork echoes the pitch directly, near Fable 5 intelligence at half the price. For a builder generating complete React Native (Expo) apps from a prompt, cheaper frontier output means more iterations per dollar and more ambitious apps within reach.
Why it fits Rork
Rork already leans on a large model roster (its AI Cloud spans 150+ models), so adding Opus 5 the day it shipped keeps its core app generation on the frontier. The practical payoff: describe a mobile app with real screens and logic, and expect both the code and the look to land closer to shippable on the first try.
Try it
Claude Opus 5 is live in Rork now. The revealing test is an app that's visual and functional at once, something with a few screens, navigation, and a real UI, and judging whether the generated result looks and works better than it did before.