Prototyping tools are not interchangeable, and picking by popularity is how teams end up doing wireframes in a high-fidelity design tool and losing a week. Sort by the question you are trying to answer instead.
Browser-based, multiplayer, and the tool most engineers already have a viewer account for. What makes it worth the price is components with variants and properties, auto layout (a constraint system close enough to flexbox that handoff maps cleanly), and variables for design tokens including modes for light and dark. Dev Mode exposes measurements, tokens and generated CSS to engineers.
Pricing is per editor per month, billed annually, with viewers free — which is the number that matters, since most of a company only ever views. Dev Mode is a separate seat type. Budget on editors, not headcount.
Penpot covers the core of design and prototyping, is free to use, and can run on your own infrastructure via Docker, which resolves the data-residency conversation that blocks SaaS design tools in regulated environments. It works natively in SVG and CSS terms, so what a designer builds maps to web primitives more directly than in most tools. It is behind Figma on plugin ecosystem and on the long tail of polish; if the blocker is licensing cost across a large team, or hosting requirements, that trade is often worth it.
Framer's distinguishing feature is that the prototype is a real website you can publish, with a CMS, responsive breakpoints and hosting included. For a marketing site or landing page, that removes the design-then-rebuild step entirely. For an application interface, it does not — you will still build the real thing. Priced per site plus per editor, so evaluate it as a site builder that designs well rather than as a design tool that publishes.
Deliberately rough hand-drawn aesthetics, free, instant, no account required. The roughness is the feature: a sketch that looks unfinished invites criticism of the idea, while a polished mockup invites criticism of the button radius. Use these for architecture diagrams, user flows and early concept work, and switch tools only once the concept is settled. Excalidraw files are plain JSON and can live in the repository next to the code they describe.
For anything involving real data, genuine latency, actual keyboard behaviour or accessibility, no design tool will tell you the truth. A throwaway app scaffolded with Vite, plus whatever component library you already use, answers questions a click-through prototype cannot: does this feel slow, does it work on a phone keyboard, what happens with 4,000 rows. The rule that keeps this useful is that it must be throwaway — the moment a prototype becomes the production codebase, you have shipped decisions you meant to test.
Not the tool — the component library. A prototype assembled from existing components takes an afternoon; one drawn from scratch takes a week and produces a design nobody can build. Invest in components with proper variants once, and every prototype after it gets cheaper. That is the same reason a design token setup pays back: change a colour in one place and every artboard follows.