How long are the videos?

Thirty seconds. That is the whole format — a single idea, one angle, one ending. It is built for shorts, ads and explainers, not long-form documentaries.

Where does the footage come from and is it legal to post?

Cosmos pulls archive pictures and film from Wikimedia Commons and credits sources on screen. Commons material is generally free to reuse, but you should verify the license on any specific asset before commercial use.

Can I edit the video after it is generated?

Not frame by frame. There is no timeline editor — you redirect by changing the brief or prompt and regenerating. It is prompt-level control, not a manual cut.

Does it make the voiceover and music too?

Yes. The pipeline writes the script, records the voiceover and generates an original score in the same run, plus finds the archive visuals that match the beats.

How does Cosmos keep facts accurate?

It researches sourced claims and ties visuals to credited Wikimedia Commons assets, with a source cited for each fact in brand films. Accuracy is bounded by what those public sources contain — treat it as a strong first draft, not a verified newsroom output.

Is there an API or bulk option for teams?

Not publicly. Volume and channel workflows go through hello@meetcosmos.com. If you need programmatic generation at scale, this is currently the wrong tool.

Directory/Video Generation/AI Video Creation/Cosmos

Cosmos

Cosmos turns a single typed idea into a finished 30-second video essay. It interviews you on angle, audience, tone and ending, then researches sourced facts, pulls real archive images and film from Wikimedia Commons (credited on screen), writes the script, records the voiceover and scores it with original music. Output includes explainers, ads, shorts, PSAs, motion graphics and music videos built from prompt-driven clips. The workflow is prompt-to-preview-to-purchase: films are free to generate and watch watermarked (up to three per day, no credit card), and you pay $9 once per film to get a clean 1080p MP4 with no subscription. Built for solo creators, social teams, educators and small brands shipping short-form video at volume without an editor or stock-footage budget.

Video Generation

Visit website

Added on

Nov 4, 2026

30-second filmsArchive footageAI voiceoverOriginal score1080p MP4Sourced factsPay per filmNo subscription

Product information

Everything worth knowing about Cosmos

The complete picture — from everyday features to the technical detail builders and IT folks dig for.

Core features

What it actually does

Idea to 30-second video essay

Type a single idea and Cosmos returns a finished 30-second film — no timeline, no editing, no stock-footage hunting.

Prompt interview before it renders

Before generating, Cosmos talks the idea through with you: the angle, who it's for, the tone, and how the film should end.

Real archive footage, credited on screen

Pictures and film are pulled from Wikimedia Commons archives and attributed on screen, not hallucinated.

Voiceover and original score included

Cosmos writes the script, records the narration and scores the film with original music — all in the same pass.

Free to make, $9 to keep

Every film is free to generate and watch watermarked, up to 3 a day. Pay $9 once per film to download the clean 1080p MP4. No subscription.

Technical capabilities

Agentic production pipeline

Research, scripting, voiceover, scoring and editing run as separate automated stages chained off one prompt rather than a single text-to-video call.

Sourced fact research layer

Facts are retrieved and attributed before the script is written, so claims carry an on-screen source instead of relying on model recall.

Must watch videos

2:21

Tutorial2:21

Cosmos AI Explained

See how Cosmos AI pairs agentic AI with Bishop Fox consultants to test applications at scale. Cosmos AI expands coverage and speed, humans verify every finding for accuracy and real-world impact. No scanner noise, just confirmed vulnerabilities your team can act on.

NVIDIA Cosmos: A World Foundation Model Platform for Physical AI

2:22

Tutorial2:22

NVIDIA Cosmos: A World Foundation Model Platform for Physical AI

The next frontier of AI is physical AI. NVIDIA Cosmos™—a platform of state-of-the-art generative world foundation models, advanced tokenizers, guardrails, and an accelerated data processing and curation pipeline—accelerates the development of physical-AI-embodied systems such as robots and autonomous vehicles. The integration of Cosmos with NVIDIA Omniverse™ creates a sim-to-real synthetic data and multiverse simulation engine for physical AI. This enables the generation of every possible outcome for a given use case, providing physical AI systems with the foresight needed to make more accurate and informed decisions. 📖 Read the Press Release → https://nvda.ws/4qeybDK 💻 Learn More About Cosmos → https://nvda.ws/49bLvCW 💻 Learn More About Omniverse → https://nvda.ws/44sK946 🧑🏻‍🍳Read the Cosmos Cookbook → https://nvda.ws/4qevli8 #NVIDIACosmos #NVIDIAOmniverse #SyntheticDataGeneration #OpenUSD #CES2025

0:56

Tutorial0:56

¡Cosmos AI evolucionó! 🚀

Te presentamos una nueva experiencia que lleva todo lo que ya conoces de Cosmos mucho más lejos. Entra a tu cuenta y descubre todo lo nuevo que puedes hacer. ✨ #CosmosAI #CosmosV2 #InteligenciaArtificial

Who uses it

Real use cases, no hype

History Explainers

Turn a single fact into a sourced 30-second film with real archival footage and on-screen credits.

Social Shorts

Fill a TikTok or Reels calendar with punchy, fact-backed videos made in minutes, not days.

Data Storytelling

Show one chart — Arctic sea ice, penny costs, fire deaths — and land a single clear takeaway.

Ad and PSA Concepts

Storyboard a campaign idea and get a watchable rough cut before committing budget.

Brand Films

Generate self-promo pieces from product screenshots or an original script you brief in chat.

Classroom and Training Clips

Produce short, accurate explainers for lessons without a video team or stock-footage license.

Motion Graphics Experiments

Let an agent write motion as code, sync it to music beats, and check the render.

What's great

Free to make and preview — 3 films/day, no card
Real archive footage from Wikimedia Commons, credited on screen
Pay $9 only for films you keep — no subscription
Sourced facts baked into every script

Technical strengths

Conversational brief capture: audience, angle, tone, ending
Single pass produces script, voiceover and original score
Clean 1080p MP4 unlocked at purchase, watermark stripped

Where it falls short

Locked to 30-second essays — no long-form output
$9 per film stacks up fast for volume work
Watermarked until you pay
Archive-only visuals — no original generative B-roll

Technical limitations

No timeline editor; you steer by prompt, not frame
Fact depth capped by what Commons and public datasets hold
No public API or bulk pipeline — email for team volume

For technical folks

The deep-dive specs

Output format

1080p MP4 download

Film length

30 seconds

Footage source

Wikimedia Commons archive pictures and film, credited on screen

Production scope

Research, script, voiceover and original music score generated in one run

FAQ

Includes technical Q&A

Making and watching films is free, up to 3 a day with a watermark. You pay $9 once per film to download the clean 1080p MP4. No subscription, no credit card to start.