Who writes here

How we work

PromptForge covers a field that changes monthly. Model lineups shift, prices are cut, capabilities appear and quietly regress. Writing about it honestly means being explicit about what we have tested, what we are repeating from a provider's documentation, and what is an opinion.

Every prompt in the library has been run against the model it names. Every tool on this site was built to solve a problem we hit ourselves. Where we describe how a model behaves, that comes from using it — not from a benchmark table. Where we quote a price or a context limit, it comes from the provider's own documentation and carries the date we checked it, because those numbers go out of date faster than anything else on the site.

We publish benchmark scores sparingly and reluctantly. They move between checkpoints, get quoted without their conditions, and give a false sense of precision about something that is genuinely hard to measure. Claims about how a model behaves in ordinary use stay true longer, and a reader can check them against their own experience — which is the point.

Corrections

When something here is wrong, we change it and say so rather than editing quietly. If you find an error — a stale price, a model that no longer behaves as described, a prompt that stopped working — write to contact@promptforge.academy. Corrections from readers are the fastest way this site stays accurate, and they are taken seriously.

Independence

PromptForge is not affiliated with OpenAI, Anthropic, Google, Meta, Mistral or any other model provider. Nobody pays for a favourable write-up, and no provider sees a page before it is published. Where we say a model is the wrong choice for something, that is the honest assessment — a comparison site that never says a product is unsuitable is not comparing anything.