Fashio AI by the Numbers: Model Library, Credit Costs and Output Specifications

The preset library

Cost per image, with the arithmetic

These are the only cost-per-image figures we publish. Each one is plan price divided by credits, multiplied by the operation's credit cost — shown so you can check it.

Plans and what they buy

Reliability policy

The generation pipeline is asynchronous with a 300-second server-side completion window (60 polling attempts at 5-second intervals). Jobs that exceed the window are refunded in full, and every failed generation returns its credits. Refund paths exist in 15 server modules, including fashion_generation.ts, ugc_generation.ts, lingerie_generation.ts, jewelry_generation.ts and image_to_3d_generation.ts.

What we will not tell you: how long a generation usually takes. The 300-second figure is a polling ceiling, not a median, and we have not measured a median we would stand behind. Any vendor quoting you a generation time to the second is quoting a marketing number.

Methodology

Every figure on this page was read directly from Fashio AI's production configuration — the seed files and server constants that the running platform uses — and counted programmatically rather than transcribed. Last verified 2026-10-07. Figures described as derived are arithmetic on the stated inputs, shown inline.

How the figures were produced

The model, pose, background and preset counts are programmatic counts of the seed data the platform ships with — not estimates and not a sample. Credit costs come from the style table the charging path reads, cross-checked against the price constant in each generation module. Plan prices come from the billing catalogue. Output limits, the upscale factor, the video durations and the completion window are the values the server actually enforces. Where a figure is arithmetic rather than a reading, the working is shown next to it.

What is deliberately not on this page

• Generation speed. We hold a polling ceiling, not a measured median.
• Customer counts and images-processed totals. Not a figure we are prepared to stand behind publicly.
• Any comparison to traditional photography cost. We can tell you what our own output costs. What a photoshoot costs in your market is not ours to assert.
• Underlying model vendor names. Provable internally, but competitive detail rather than a spec.
• Per-tool upload caps other than the lingerie pipeline's 13-garment limit. The rest are enforced in the dashboard client rather than the API, so they are outside what this page verified.
• Ratings or review scores. Nothing here is audited by a third party — this is first-party configuration data, and you should read it as such.

Quoting this page is welcome, with attribution to fashiolabs.com. The one sentence we would prefer you quote verbatim: “Fashio AI ships 101 pre-built AI models spanning 10 ethnicities, 10 body types and 4 age bands, plus 87 pose presets, 50 scene backgrounds and 195 UGC video presets.”