Owen Pryce

Totalum13 August 2026

Backwards in the right way, for the right projects

Totalum asks you to define the data model before you see a screen, which our non-technical testers hated and our second brief rewarded. Brief two is an inventory tool with staff administration; Totalum was the only entrant where the admin side simply existed once the tables did, rather than being another two days of CRUD screens. It also has the second strongest API and MCP surface here at 4.4, behind Replit alone, which matters if you plan to script against your own product. Marked down on design: the front end is competent and unmemorable.

Mocha13 August 2026

Cheap in the way that actually counts

I track real spend on every build in this index, including credits burned on failed attempts. Mocha finished our two small briefs for under 6 EUR of usage each, and it was the only tool where the cost of a failed attempt was low enough that nobody hesitated to retry. That is what a 4.1 value score means in practice. It also has no public API, no roles model and a scalability ceiling that arrives early, which is why it does not rank higher overall.

Fair request. Per-brief spend is in the testing spec on our methodology page and we are working on publishing the raw logs.