About ShortsFast

Every frontier AI video and image model under one $20/month subscription. Full commercial license. No watermarks on paid output. No tiered chaos.

What this blog is for

blog.shortsfa.st is the documentation surface for people picking AI video and image models. Every post answers a single decision: which model to pick, how to prompt it, what it costs, and where it breaks. We test the models on the platform we ship — shortsfa.st — so the recipes you read here reproduce on the same generation pipeline you'd run yourself.

Who writes here

Posts are bylined ShortsFast Team — a collective editorial byline used when a piece is the product of multiple people on staff (research, prompt testing, drafting, technical review). Where a single named author makes sense, we name them. Every published post lists a reviewedBy field; on solo posts the writer and reviewer are the same person, marked explicitly.

Editorial standards

Five non-negotiables apply to every post we publish:

  1. Answer-block opener. Every post leads with a 40-70 word direct answer to the question the title asks. No throat-clearing, no "in this article we will" preamble.
  2. Author + reviewer named. Both the writer and the reviewer are listed in frontmatter and rendered on the post. Solo posts mark writer = reviewer.
  3. Last-updated date emitted. Every post sets an updatedAt frontmatter date that emits as dateModified in the post's BlogPosting JSON-LD. We never bump the date without a substantive content change — Google calls that out as a spam signal and we agree.
  4. ≥3 primary-source citations. Vendor docs, model cards, papers, or official release notes. Third-party recaps don't count as primary. Each citation links the canonical source URL, never an aggregator.
  5. Sources block at the bottom. A visible list of every source cited in the post, with the date we read it. If a source moves or disappears, the next refresh updates the link or replaces the citation.

How we test models

We run every prompt we recommend on the live ShortsFast generation pipeline. For comparison posts (e.g. FLUX 2 Pro vs Nano Banana Pro) we use identical prompts, seeds, and parameters across each candidate model and capture the unedited first-pass output. We don't cherry-pick the best of N runs unless we say so explicitly. When a recipe in a post can't be reproduced on the current model version, we either update the recipe or pull the post.

Pricing claims are verified against the model vendor's public pricing page on the day the post is published or refreshed. When we say "12× cheaper," we show the multiplication.

Corrections

Spotted something wrong? Email hello@shortsfa.st with the post URL and the issue. We log every correction in the post's updatedAt field and leave a note in the source-list block when a citation is replaced.

Disclosure

ShortsFast is a paid product ($20/month). Posts that demonstrate a recipe link to the prompt pack page on this blog and to the product surface on shortsfa.st. We don't run paid placements, sponsored posts, or affiliate links for competitors — when we cite a competitor (Runway, Pika, Luma, etc.) we link the vendor's own page, never an affiliate URL.

Contact