FASHN
Proprietary virtual try-on API — swap garments onto any model photo with photorealistic accuracy in seconds.
Overview
FASHN (Tel Aviv, founded 2022) trained its own 1.2B-parameter pixel-space try-on model from scratch rather than wrapping a foundation model, offered via web studio or API (Python/TypeScript SDKs) from $0.075/credit down to about $0.05 at the top commitment tier.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 11 Jul 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- Product-to-Model
- Model Swap
- Try-On
- Basic AI editing
- 200 credits/mo
- 4K image gen & upscaling
- 720p AI video
- 2 team seats
- 750 credits/mo + 50/day
- 1080p AI video
- 5 team seats
- Priority support
- 1,500 credits/mo + 100/day
- Custom face reference uploads
- 10 team seats
- Priority feature requests
Use cases
What you can produce with FASHN
- Photorealistic try-on image (garment composited onto a person photo) in 5-17 seconds
- On-model product shot generated from a flat-lay or ghost mannequin photograph
- AI-generated fashion model image with customizable demographics and styling
- Short motion video clip generated from a single fashion image (resolution varies by plan tier)
- Background-removed product cutout with transparent background
- Reframed or cropped image with intelligent aspect-ratio adjustment
- Commercial-use license included with all API outputs at no extra charge
ASEAN Perspective
FASHN in Southeast Asia
FASHN's API is cloud-hosted and globally accessible, making it technically available to ASEAN fashion e-commerce brands in Singapore, Malaysia, Indonesia, and Vietnam without any regional restrictions. The APAC fashion AI market is growing at over 40% annually and virtual try-on adoption is accelerating across platforms like Lazada and Shopee, creating natural demand for tools like FASHN. However, FASHN has not publicly documented ASEAN clients, local partnerships, or regional pricing in SGD/MYR/IDR, and its documentation and support appear primarily English-language. For ASEAN brands with development teams, the per-image API economics are competitive; for those without technical resources, the absence of a plug-and-play integration is a gap.
FASHN is a technically credible virtual try-on API that stands apart from the crowded field of wrapper tools. Its proprietary pixel-space model, trained on 18 million examples, delivers strong garment detail preservation — especially on printed fabrics and complex textures — and the credit-based pricing makes it one of the most cost-efficient options for high-volume e-commerce catalog workflows. The January 2026 open-sourcing of VTON v1.5 signals genuine research depth for a bootstrapped two-person team, and the Python/TypeScript SDKs lower the integration barrier for developers.
The caveats are meaningful. The base output resolution (576x864 on v1.5) is lower than some competitors, and higher-fidelity outputs cost more credits. The platform is developer-first with no native Shopify or plug-and-play integration, creating friction for non-technical fashion teams. There is no permanent free tier beyond 10 trial credits. APAC adoption is not documented, and the company's tiny team size introduces delivery-risk if demand scales sharply. Worth evaluating carefully against Genlook and Botika for head-to-head resolution and pose-control comparisons before committing.
What people say
Most virtual try-on startups wrap someone else's diffusion model and call it a product. FASHN didn't — the Tel Aviv team (founded 2022, self-funded) trained its own 1.2-billion-parameter flow-matching model from scratch on roughly 4 million fashion images, running directly in pixel space instead of a compressed latent, which is why texture and pattern detail hold up better than most competitors on printed or metallic fabric.
The API covers the practical range: tops, bottoms, dresses, jackets, front/back/three-quarter poses, maskless inference so it doesn't need a segmentation step, and outputs in 5-17 seconds. Beyond try-on there's Product-to-Model, Face-to-Model, model generation and swapping, background removal, and image-to-video, all reachable through a REST API with Python and TypeScript SDKs, or a simpler web studio for smaller teams. Base output on the current open-sourced model (v1.5, Apache-2.0 since January 2026) sits at 576x864 — usable, but lower than some rivals ship by default.
Pricing runs on credits: $0.075 per credit with no commitment, dropping toward $0.05 at the highest-volume tier. The app side has its own subscription ladder — Free (10 credits), Basic ($19/mo), Pro ($49/mo, the most-picked plan), Agency ($99/mo) — separate from the API's usage-based pricing, which can be confusing to compare at a glance. There's no Shopify plug-in or similar no-code integration, so non-technical teams need a developer to wire it in. Independent review volume is still thin for a two-to-small-team operation, so treat any specific accuracy claims as directional rather than statistically proven — worth comparing head-to-head against Genlook or Botika before committing budget.
Summary of public user & expert reviews, compiled by RECATOOLS.
Notable facts
- FASHN was built by a husband-and-wife team working out of their living room in Belgrade, Serbia — inspired partly by watching the K-Drama 'Start-Up!' during the pandemic.
- The company remains fully bootstrapped and self-funded, reaching $25,000+ in monthly recurring revenue without venture capital.
- FASHN's v1.5 model operates in pixel space rather than latent space — an unusual architectural choice that preserves fine fabric texture and printed details more faithfully than VAE-based diffusion models.
- FASHN open-sourced its VTON v1.5 model weights on Hugging Face under Apache-2.0 in January 2026, allowing researchers to fine-tune the 972M-parameter model on consumer GPUs for as little as $5,000–$10,000.
Frequently asked questions
About this listing
This entry was compiled from publicly available data including FASHN's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with FASHN unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to FASHN directly →
Spotted something out of date? Suggest an update →
FASHN in the news
Alternatives to FASHN
More in Image Generation
