VeizikNVIDIA, Apple Silicon and edge builds are available now.Release notes →
veiziknative inference engine
For business Sign inDownload
← Engineering
Ease of use

From a prompt to a video, in one session.

What the product feels like to use: a four-command start, and three models — language, image and video — chained on one engine. The measured end-to-end timing follows; the shape and the output are here now.

AI-generated · the chain’s output — WAN 2.2 i2v

AI-generated · Powered by Wan-AI/Alibaba · the final stage of the chain below

Four commands to a render

Install, activate, pull, run.

After the one-time download, your first render is one command.

1
veizik install flux

Fetch the model engine and verify its SHA-256. One package.

2
veizik activate

Approve this machine once; the licence binds to the device.

3
veizik pull flux1schnell

Fetch the checkpoint into the local cache, ready to run.

4
veizik run flux1schnell "a lighthouse at dawn"

Generate once and exit — your first render.

One session, three models

Language writes it, image draws it, video moves it.

The cross-modal chain on one engine, counted end-to-end.

  • One session, three models. A language model writes a prompt, an image model draws it, and WAN 2.2 image-to-video animates the still — language, image and video on the same native engine, switched without leaving the session.
  • The whole chain on one timeline. The schedule below counts every part — the model switch, the load and the render at each stage — as one continuous run, so the number is the experience, not a best-case for one step.
  • The live, fastest run is coming. The measured end-to-end timing — with the fastest path on the hypercache build — is being prepared; this page carries the shape of the chain and its output now, and the figures follow.
ONE SESSION · THREE MODELS01Languagewrites the promptQwenswitch02Imagedraws from the promptFLUX.1 / SD3.5switch03Videoanimates the stillWAN 2.2 i2vper stage: model switch → load → render — one continuous timelinemeasured end-to-end run — coming on the hypercache build
A tranquil alpine lake at sunrise with mirror reflections of snow-capped peaks and a pine-forest shoreline — AI-generated with FLUX.1-schnell on veizik (NVIDIA RTX 3090).
FLUX.1-schnellNVIDIA RTX 3090

The image stage, shown: FLUX.1-schnell drawn natively by Veizik on an RTX 3090 — output, not a benchmark.