Framework choices affect AI coding cost

A benchmark comparing AI-built JavaScript apps found that framework choice changed the cost of each run, but not by a huge amount in this test. Hono with htmx had the lowest median at $2.67, while Angular had the highest at $4.97. The author ran the same small Pokémon app build several times per framework, using an AI coding agent, then measured cost alongside time, code size, performance, and accessibility.

Treat those numbers as a comparison, not a price list for your next software project. The test covered one relatively simple, content-focused app, and the author says results varied between runs because both the AI agent and Lighthouse can behave inconsistently. The costs are quoted at API prices, not what the author paid, since the runs were part of a monthly plan. The post drew 232 points and 164 comments on Hacker News, a sign that developers are paying attention to the practical cost of framework choices. Read the benchmark and its caveats.

For a business, the useful lesson is to test your own work before assuming one framework will make AI coding cheaper. Cost per run is one input, but the benchmark also looked at whether the resulting app met its spec and how it performed. This is the kind of comparison teams can use to make choices using evidence rather than assumptions.