Qurak

Compared with the Wolfram Language

The same syntax, a different engine. Here is exactly how much of the reference runs, and where it does not.

The numbers

Of the 6,300 symbols in the .wl reference that Qurak documents, 2,834 run here with the examples shown executed, 130 run here and are known to answer differently, 44 are implemented by the engine and refused by this deployment's policy, and 3,292 are not implemented at all.

Counted from the reference at request time, not typed here. The dataset the counts come from is linked below.

What is the same

The language. Expressions you have written for the Wolfram Language - the bracket syntax, the capitalised built-ins, Rule and ReplaceAll, the list conventions - are what Qurak evaluates. Calculus, linear algebra, series, differential equations, number theory, statistics and plotting are the core, and the core is where the {supported} live.

What is different by design

Qurak is a hosted service with an HTTP API and an MCP server on every plan, including the free one; there is nothing to install and nothing to license per seat. Every evaluation runs in a confined worker process. The reference states availability per symbol and every example on it was executed by this engine - a promise the incumbent does not make about its documentation, and one we can only make because we say no in public.

Where Qurak loses

3,292 symbols are not implemented, and they include whole areas - image processing, geographic computation, machine learning, the knowledge base of curated data, most of the notebook interface. 130 symbols answer differently from the reference, each documented on its page. Some integrals return unevaluated that Mathematica would close. If your work lives in those areas, Qurak is not the tool, and this page exists to tell you so before you find out.

Where Qurak wins

Price and access: the free plan has no card and no expiry, and paid plans buy compute budget and rate rather than permission. An AI assistant can call it directly over MCP and get a computed answer instead of a plausible one. And the honesty is the product: you know before you rely on a function whether it runs here.

Wolfram Language and Mathematica are trademarks of Wolfram Research, which is not affiliated with Qurak. The comparison is of what runs, counted; it is not a claim about quality where both run.

What runs here, and what does not · availability.json