Viral AI Trends
Last updated

Every new AI video trend, explained and made free

A new AI video format goes viral roughly every few weeks, and the model behind it changes just as often. Each trend gets one page answering the same three questions: what exactly is it, how do I make one, and is there a way to do it without paying. The models get their own pages too — priced, dated and read from the makers' own documentation rather than from launch coverage.

Trends covered

Each entry below is a complete guide: what the format is, where it came from, how to make one from your own photos, the prompts that work, and which tools are free.

AI 80s trend Rising Turn a modern selfie into a 1986 portrait, then animate it into a VHS-style clip. The two-step workflow, eight prompts that hold the face, the free routes — and what happens to the photo you upload. Hotel Lobby AI Two photos, one orange booth, one hanging microphone. The duet format that took over feeds from mid-September 2026 — the full walkthrough, eight prompts, the model behind it and the permission rules. Muse AI (Meta) New Meta's personal AI agent runs on its own cloud machine and does the task instead of describing it. What it is, the invite-only signup rules, the 1 billion token mechanic, pricing — and the US-only regional limit. Image to video AI New Every kind of image-to-video generator in one comparison: what free actually gives you, what each type reuses, the four-step workflow, and the prompt pattern that stops faces morphing.

Models, priced from their own sources

Model pages are built the other way round: from the maker's model card, pricing page and launch material, with the route in front of the number. What it costs at every resolution, what changed from the last version, what is genuinely free, and what nobody has published.

Gemini Omni New Google's any-to-any video model, and the one it built instead of a Veo 4. The free YouTube route, the token maths behind the $0.10-a-second headline, the 10-second cap — and the two leaderboards that now disagree about it. Nano Banana 2.1 New Google halved the price of image generation on 6 October 2026. What every resolution costs across the whole family, the honest split between the benchmark and the picture when you compare it with Pro, and a 29 October migration deadline. GPT Image 2.5 New Two models at one price, with no cost reason to pick one. The real difference between Flare and Sunburst, the 4× cheaper high-quality tier nobody announced, native transparent backgrounds, and the free ChatGPT route. MiniMax H3 New Ranked first on Arena's blind image-to-video board — and one of the few frontier video models you can download. H3 versus H3 Max, every per-second rate, and the two licence conditions that catch commercial users. Seedance 2.5 ByteDance Seed's 30-second video model, read from primary sources: what changed from 2.0, what Dreamina, the BytePlus API and aggregators charge, the copyright settlement behind it — and the four specifications nobody has published. Kling 4.0 Status page Announced but not released. What Kuaishou has actually stated, what is still only reported, what Kling 4.0 Flash gives you today, and what Kling really costs once you price credits properly. Grok Imagine New The free tier came back in October 2026 with a hard cap of one 480p clip a day. Every per-second API rate, the three video models, the capability the newest one dropped, and why the company that builds it has changed its name twice. Higgsfield AI The model aggregator reviewed from its own pricing page: what each plan really grants, why credits vanish faster than the allowance suggests, the four surfaces that always bill, and the 7-day zero-usage refund rule.

Where you actually go to use these

A guide that explains what a model is but never says where to use it has only done half the job. Every model on this site has an official home, and this is it. These are the vendors' own pages — we are not affiliated with any of them and take no payment for listing them.

ModelWhere to use it — official
Seedance 2.5 Dreamina · Jimeng (China) · BytePlus API
Wan 3.0 Wanxiang · Alibaba Cloud Bailian · Wan-AI weights (2.x only)
Higgsfield Genjutsu higgsfield.ai/genjutsu
Kling 4.0 Kling AI · Kling (China)
MiniMax H3 Hailuo AI · Hugging Face weights
Grok Imagine grok.com/imagine
Gemini Omni Gemini · Google AI Studio
Nano Banana 2.1 Google AI Studio · Gemini
GPT Image 2.5 ChatGPT · OpenAI API docs
Higgsfield higgsfield.ai
Muse (Meta) Meta newsroom announcement

Links are to the vendors' own sites and open in a new tab. Prices and availability change fast — always confirm on the page you are buying from. Every model page on this site repeats its own official links at the top.

Reference

Free AI video tools The generator comparison that stays the same as trends change: what each approach costs, what it reuses, and what to check before you pay. AI video tool status tracker Updated Which AI video tools still serve a working product and which quietly stopped. Every status comes from a direct request to the tool's own domain, with the date it was checked.

How this site works

Most format guides are written once and abandoned, so they go stale exactly when the format is still climbing. This site is built the other way round:

What these trends have in common

Strip away the specifics and most of these formats work the same way. Something with a highly recognisable look appears in music videos, film or advertising. Someone works out how to reproduce that look with an image-to-video model using ordinary photos. The result spreads because the format announces the joke before anyone reads a caption — the payoff comes from who you put in the scene, not from editing skill.

That is also why they fade: the staging dates, while the capability underneath does not. Learning the workflow once is worth more than chasing any single one of them.

Common questions

What is Viral AI Trends?

An independent, English-language guide to AI video formats as they go viral. Each format gets one page answering the same three questions: what it is, how to make one from your own photos, and which routes are genuinely free.

How do you decide which formats get a page?

When a format starts climbing in search and social feeds, we research it, check whether anyone has already written a proper guide, and publish one if there is a gap. New pages are announced in the RSS feed.

Do I have to pay to make these videos?

Usually not for your first one. Most formats have a genuinely free route for the photo or template step, while the higher-fidelity video models sit behind paid plans. Each guide lists the free routes explicitly, with prices dated so you know how fresh they are.

Is it safe to upload photos of myself?

It depends on the tool's data settings, and every guide has a section on exactly this: what the major platforms do with uploads by default, how to switch training use off, and which photos you should never upload. See the guide you are using for the specifics.

Can I make one using someone else's face?

Only with their permission. Our guides state this plainly: transforming images of non-consenting people — including celebrities — causes harm and gets posts removed, and we do not provide prompts for it.

Disclosure. This site is an independent guide. It is not affiliated with, endorsed by or sponsored by any artist, production company or tool listed. Some links may be affiliate links; they do not change what we recommend. Prices and features change quickly — always confirm on the official page.