One SDK for Kling, Flux, ElevenLabs and more. Built on Vercel AI SDK.
Docs · Dashboard · Quickstart · Models and prices · How to · Discord
Warning
JSX/TSX videos are deprecated. vargai/react (<Render>, <Clip>, <Video>…), vargai/studio and the JSX runtime still work but get no new features, and render() prints a DeprecationWarning (code VARGAI_JSX). The CLI no longer renders: vargai render, preview, frame, storyboard, studio and hello were removed.
Generate with the varg API or the varg MCP tools and edit with the ffmpeg tool, from the terminal:
curl -fsSL https://varg.ai/install | sh
vargai login
vargai generate_video --prompt "waves at dusk" --duration 5
vargai get_tool ffmpeg # the ffmpeg tool's input, for call_toolThe rest of this README describes the deprecated JSX API.
varg is an open-source TypeScript SDK for AI video generation. One API key, one API — generate images, video, speech, music, lipsync, and captions through varg.* providers. Write videos as JSX components (like React), render locally or in the cloud.
Install the varg skill into Claude Code, Cursor, Windsurf, or any agent that supports skills. Zero code — just prompt.
# 1. Install the varg skill
npx -y skills add vargHQ/skills --all --copy -y
# 2. Set your API key (get one at varg.ai/dashboard/api; a new account gets 50 free credits)
export VARG_API_KEY=varg_live_xxx
# 3. Create your first video
claude "create a 10-second product video for white sneakers, 9:16, UGC style, with captions and background music"The agent writes declarative JSX, varg handles AI generation + caching + rendering.
# Install with bun (recommended)
bun install vargai ai
# Or with npm
npm install vargai ai
# Set up project (auth, skills, cache dirs)
bunx vargai initvargai init signs you in, installs the agent skill and sets up your project structure.
Or ask your AI agent to create something from scratch.
Your prompt / JSX code
|
varg API (api.varg.ai/v2)
/ | \ \
Kling Flux ElevenLabs Wan ... (AI providers)
\ | / /
varg render engine
|
output.mp4
- One API key (
VARG_API_KEY) routes to all providers through the varg API - Declarative JSX — compose videos like React components with
<Clip>,<Video>,<Music>,<Captions> - Automatic caching — same props = instant cache hit at $0. Re-render without re-generating
- Local or cloud — render with
render()fromvargai/reactlocally, or submit via the cloud render endpoint (POST https://api.varg.ai/v2/render)
/** @jsxImportSource vargai */
import { Render, Clip, Image, Video } from "vargai/react";
import { varg } from "vargai/ai";
const character = Image({
prompt: "cute kawaii orange cat, round body, big eyes, Pixar style",
model: varg.imageModel("nano-banana-pro"),
aspectRatio: "9:16",
});
export default (
<Render width={1080} height={1920}>
<Clip duration={5}>
<Video
prompt={{ text: "cat waves hello, bounces happily", images: [character] }}
model={varg.videoModel("kling-v3")}
/>
</Clip>
</Render>
);// render.ts — the CLI no longer renders, so call render() yourself.
// hello.tsx needs the JSX pragma on its first line (see above).
import { render } from "vargai/react";
import video from "./hello.tsx";
await render(video, { output: "output/hello.mp4" });import { Render, Clip, Image, Video, Speech, Captions, Music } from "vargai/react";
import { varg } from "vargai/ai";
const character = Image({
model: varg.imageModel("nano-banana-pro"),
prompt: "friendly robot, blue metallic, expressive eyes",
aspectRatio: "9:16",
});
const voiceover = Speech({
model: varg.speechModel("eleven_v3"),
voice: "adam",
children: "Hello! I'm your AI assistant. Let me show you something cool!",
});
export default (
<Render width={1080} height={1920}>
<Music prompt="upbeat electronic, cheerful" model={varg.musicModel()} volume={0.15} />
<Clip duration={5}>
<Video
prompt={{ text: "robot talking, subtle head movements", images: [character] }}
model={varg.videoModel("kling-v3")}
/>
</Clip>
<Captions src={voiceover} style="tiktok" color="#ffffff" withAudio />
</Render>
);import { Render, Clip, Image, Video, Speech, Captions, Music } from "vargai/react";
import { varg } from "vargai/ai";
const voiceover = Speech({
model: varg.speechModel("eleven_v3"),
voice: "josh",
children: "With varg, you can create any videos at scale!",
});
const baseCharacter = Image({
prompt: "woman, sleek black bob hair, fitted black t-shirt, natural look",
model: varg.imageModel("nano-banana-pro"),
aspectRatio: "9:16",
});
const animatedCharacter = Video({
prompt: {
text: "woman speaking naturally, subtle head movements, friendly expression",
images: [baseCharacter],
},
model: varg.videoModel("kling-v3"),
});
export default (
<Render width={1080} height={1920}>
<Music prompt="modern tech ambient, subtle electronic" model={varg.musicModel()} volume={0.1} />
<Clip duration={5}>
<Video
prompt={{ video: animatedCharacter, audio: voiceover }}
model={varg.videoModel("sync-v2-pro")}
/>
</Clip>
<Captions src={voiceover} style="tiktok" color="#ffffff" withAudio />
</Render>
);| Component | Purpose | Key props |
|---|---|---|
<Render> |
Root container | width, height, fps |
<Clip> |
Time segment | duration, transition, cutFrom, cutTo |
<Image> |
AI or static image | prompt, src, model, zoom, aspectRatio, resize |
<Video> |
AI or source video | prompt, src, model, volume, cutFrom, cutTo |
<Speech> |
Text-to-speech | voice, model, volume, children |
<Music> |
Background music | prompt, src, model, volume, loop, ducking |
<Title> |
Text overlay | position, color, start, end |
<Subtitle> |
Subtitle text | backgroundColor |
<Captions> |
Auto-generated subs | src, srt, style, color, activeColor, withAudio |
<Overlay> |
Positioned layer | left, top, width, height, keepAudio |
<Split> |
Side-by-side | direction |
<Slider> |
Before/after reveal | direction |
<Swipe> |
Tinder-style cards | direction, interval |
<TalkingHead> |
Animated character | character, src, voice, model, lipsyncModel |
<Packshot> |
End card with CTA | background, logo, cta, blinkCta |
<Captions src={voiceover} style="tiktok" /> // word-by-word highlight
<Captions src={voiceover} style="karaoke" /> // fill left-to-right
<Captions src={voiceover} style="bounce" /> // words bounce in
<Captions src={voiceover} style="typewriter" /> // typing effect67 GL transitions available:
<Clip transition={{ name: "fade", duration: 0.5 }}>
<Clip transition={{ name: "crossfade", duration: 0.5 }}>
<Clip transition={{ name: "wipeleft", duration: 0.5 }}>
<Clip transition={{ name: "cube", duration: 0.8 }}>All models are accessed through varg.* — one API key, one provider.
import { varg } from "vargai/ai";| Model | Use case | Credits (5s) |
|---|---|---|
varg.videoModel("kling-v3") |
Best quality, latest | 150 |
varg.videoModel("kling-v3-standard") |
Good quality, cheaper | 50 |
varg.videoModel("kling-v2.5") |
Previous gen, reliable | 50 |
varg.videoModel("wan-2.5") |
Good for characters | 50 |
varg.videoModel("minimax") |
Alternative | 50 |
varg.videoModel("sync-v2-pro") |
Lipsync (video + audio) | 50 |
| Model | Use case | Credits |
|---|---|---|
varg.imageModel("nano-banana-pro") |
Versatile, fast | 5 |
varg.imageModel("nano-banana-pro/edit") |
Image-to-image editing | 5 |
varg.imageModel("flux-schnell") |
Fast generation | 5 |
varg.imageModel("flux-pro") |
High quality | 25 |
varg.imageModel("recraft-v3") |
Alternative | 10 |
| Model | Use case | Credits |
|---|---|---|
varg.speechModel("eleven_v3") |
Text-to-speech | 25 |
varg.speechModel("eleven_multilingual_v2") |
Multilingual TTS | 25 |
varg.musicModel() |
Music generation | 25 |
varg.transcriptionModel("whisper") |
Speech-to-text | 5 |
1 credit = $0.01. Cache hits are always free.
bunx vargai login # sign in (email OTP or API key)
bunx vargai init # set up project (auth + skills)
bunx vargai balance # check credit balance
bunx vargai topup # add credits
bunx vargai run image --prompt "sunset" # generate a single image
bunx vargai run video --prompt "ocean waves" # generate a single video
bunx vargai list # list available models and actionsEverything the varg MCP server does is a command too, tool for tool. The standalone vargai (one file, no Node or Bun) installs with:
curl -fsSL https://varg.ai/install | sh # macOS, Linux
irm https://varg.ai/install.ps1 | iex # Windows
vargai update # later: the latest releasevargai search "captions for reels" # varg how to: skills, guides, presets (no key)
vargai install remotion-captions # install a skill into your agent (no key)
vargai version # this version and the latest (no key)
vargai tools # every MCP tool; ○ = no key needed
vargai login # then: account, balance, every model
vargai generate_image "a red fox" # any MCP tool, options from its schema
vargai help generate_video # a tool's optionsFrom the npm package (bunx vargai …) the same commands work as vargai mcp <tool>, and they use the installed binary when there is one. The SDK commands (init, run, list…) run on Bun either way. The CLI does not render JSX/TSX any more. Releases: cli-v*.
# Required — one key for everything
VARG_API_KEY=varg_live_xxxGet your API key at varg.ai/dashboard/api. A new account gets 50 free credits ($0.50) on its first sign-in; after that you pay each model's provider price, listed at varg.ai/models. Bun auto-loads .env files.
Bring your own keys (optional)
You can use provider keys directly if you prefer:
FAL_API_KEY=fal_xxx # fal.ai direct
ELEVENLABS_API_KEY=xxx # ElevenLabs direct
OPENAI_API_KEY=sk_xxx # OpenAI / Sora
REPLICATE_API_TOKEN=r8_xxx # Replicate
KREA_API_KEY=xxx # Krea direct (krea.imageModel("krea"))See the BYOK docs for details.
| Action | Model | Credits | Cost |
|---|---|---|---|
| Image | nano-banana-pro | 5 | $0.05 |
| Image | flux-pro | 25 | $0.25 |
| Video (5s) | kling-v3 | 150 | $1.50 |
| Speech | eleven_v3 | 25 | $0.25 |
| Music | music_v1 | 25 | $0.25 |
| Cache hit | any | 0 | $0.00 |
A typical 3-clip video costs $2-5. Cache hits are always free.
https://discord.com/channels/1463766525117861922/1463766526250188824
See CONTRIBUTING.md for development setup.
Apache-2.0 — see LICENSE.md