Build with elevideo

elevideo turns raw videos into captioned and AI-edited content through a simple REST API, CLI, or MCP integration. Here's what you can do:

• Add captions — upload any video and get back styled, word-timed captions in 30+ visual styles
• Apply AI edits — let AI analyze your video's transcript and generate overlay graphics (property listings, product demos, podcast clips, and more)
• Re-style for free — once a video is transcribed, re-run it with any other style at no extra cost

New here? Start with the Quickstart below — you'll have a captioned video in under 5 minutes.

Get your first captioned video in under 5 minutes.

1. Get an API key — create one below or on your account page
2. Caption your video — call POST /api/v1/captions with any public video_url and optional style_id. You'll get back a run ID immediately while we process it in the background.
3. Wait for it — call GET /api/v1/render/:id until status is "completed". Your captions are ready.
4. Download your video — call POST /api/v1/render/:id/export, then poll until export_status is "completed" and grab the download_url.

Here's the complete flow:

# 1. Add captions to a video
curl -X POST https://api.elevideo.dev/v1/captions \
  -H "X-API-Key: ck_your_key" \
  -H "Content-Type: application/json" \
  -d '{"video_url": "https://example.com/video.mp4", "style_id": "holographic"}'

# 2. Poll status (repeat until status = "completed")
curl https://api.elevideo.dev/v1/render/run_abc123 \
  -H "X-API-Key: ck_your_key"

# 3. Trigger export
curl -X POST https://api.elevideo.dev/v1/render/run_abc123/export \
  -H "X-API-Key: ck_your_key"

# 4. Poll again until export_status = "completed", then use download_url

# 5. Try a different style (free — no re-transcription)
curl -X POST https://api.elevideo.dev/v1/render/run_abc123/rerun \
  -H "X-API-Key: ck_your_key" \
  -H "Content-Type: application/json" \
  -d '{"style_id": "beast"}'

Once you have this working, try applying a different style — see Browse Styles for all 32 options.

The pipeline

Upload Video ──▶ POST /captions ──▶ Transcription ──▶ Composition Built
                    or POST /ai-edits                     status: "completed"
                                                                │
                                                                ▼
                                                    POST /render/:id/export
                                                         (async)
                                                                │
                                                                ▼
                                                      Server-side Render
                                                                │
                                                                ▼
                                                    Upload to CDN ──▶ download_url

1. You provide a video URL and optionally a style ID
2. The API transcribes your video with word-level timing
3. For caption styles: builds a composition with styled captions overlaid on your video
4. For AI edit styles: also uses AI to extract highlights, quotes, or structured data from the transcript
5. Once the composition is built, you trigger an export which renders a final MP4 on our servers
6. The rendered MP4 is uploaded to CDN and a download URL is returned

What's a composition?

Every completed render produces a composition object — the full set of rendering instructions including captions, timing, styles, and overlays. You can retrieve any render's composition via GET /api/v1/render/:id. It contains:

• segments — video clips with timing, transitions, animations
• captions — word-level caption data with style configuration
• transcriptionWords — raw word timing from transcription
• audioUrl — extracted audio track
• styleParams — video effects (zoom, pan, blur)
• aiEditTemplate — AI-extracted overlay data (for AI edit styles)

Every video you render is effectively a reusable template. Process the same video with different styles and get different outputs — all from a single transcription.

Tracking progress

Both render and export operations are asynchronous. The API responds immediately and processes in the background. Poll the render endpoint to track progress.

Status transitions:

processing ──▶ completed ──▶ (trigger export) ──▶ rendering ──▶ uploading ──▶ completed
     │                                                         │
     ▼                                                         ▼
   failed                                                    failed

• processing — transcription + composition in progress
• completed — composition ready, can view or export
• failed — processing error (check error_message)
• rendering — server-side export in progress
• uploading — rendered MP4 uploading to CDN
• export_status: "completed" — download_url available

Re-runs: change styles for free

Once a video is transcribed, you can re-run it with any other style at no cost using POST /render/:id/rerun. Caption re-runs complete instantly. AI edit re-runs only need a brief LLM step. No credits charged.

Now that you understand the flow, explore Add Captions or Apply AI Edits.

Prefer working from the terminal? The elevideo CLI handles uploads, rendering, and downloads in one command.

Install with npm:

npm install -g @elevideo/cli

Set your API key:

elevideo auth --key ck_your_key

Then render a video:

elevideo render ./video.mp4 --style bold --flow captions
# ✓ Uploaded  (12.4 MB)
# ✓ Rendering … done
# ✓ Exported  → video-bold.mp4

elevideo styles captions # list available caption styles
elevideo renders         # list past renders
elevideo status <run-id> # check render status

Let AI agents render videos on your behalf. The elevideo MCP server connects to Claude Desktop, Cursor, or any MCP-compatible client — your agent calls the tools, we handle the rest.

Install the server:

npm install -g @elevideo/mcp-server

Claude Desktop — add to ~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "elevideo": {
      "command": "elevideo-mcp",
      "env": {
        "ELEVIDEO_API_KEY": "ck_your_key"
      }
    }
  }
}

Cursor — add to .cursor/mcp.json in your project

{
  "mcpServers": {
    "elevideo": {
      "command": "elevideo-mcp",
      "env": {
        "ELEVIDEO_API_KEY": "ck_your_key"
      }
    }
  }
}

Available tools:

• render_video — render a video with a template
• check_render_status — poll render progress
• export_render — trigger MP4 export
• list_templates — list all caption styles and AI templates
• get_account — check credit balance

Every API request needs to be authenticated so we know who's making it and which credit balance to use. Pass your API key in the X-API-Key header:

curl -H "X-API-Key: ck_your_key_here" \
  https://api.elevideo.dev/v1/account

Keep your key private — anyone with it can use your credits. If a key is compromised, revoke it immediately from the Account & Keys section below.

Bearer token authentication (Authorization: Bearer <token>) is also supported for browser-based integrations.

Loading keys...

Turn any video into a captioned video. Send a URL and a style, and we'll transcribe the audio, generate word-timed captions, and build a composition you can export as MP4.

32 caption styles across 4 categories: minimal, bold, animated, and elegant. Each controls fonts, colors, animations, and effects. Call GET /api/v1/styles/captions for the full list.

Popular styles:

"classic"       — Traditional subtitles with dark background (minimal)
"holographic"   — Cyan holographic glow with uppercase text (animated)
"beast"         — Impact font with yellow box on active word (bold)
"elegance"      — Serif script with alternating cursive words (elegant)

Want a different look? You can re-run with a new style for free after the first render.

Transform a video with an AI-generated overlay. We transcribe the audio, then use AI to pull out the most relevant content — quotes, features, stats, or highlights — and build a styled overlay on top. Pick a style that matches your use case.

32 AI edit overlay styles organized by industry: real estate, SaaS, personal brand, trendy, podcast, news, reviews, local business, e-commerce, and finance. Each uses AI to extract relevant data from your video's transcript. Call GET /api/v1/styles/ai-edits for the full list.

Popular styles:

"property_showcase"  — Property listing with price + features (real_estate)
"product_demo"       — SaaS product walkthrough (saas)
"tiktok_title"       — Trendy short-form content (trendy)
"podcast_clip"       — Interview clips with standout quote (podcast)

Browse all available visual styles before starting a render. Each style has a unique ID you'll pass to the captions or AI edits endpoint.

Once your composition is ready, trigger an export to produce a downloadable MP4. The export runs on our servers and typically takes 1–8 minutes depending on video length.

Try a different style without re-processing your video. Since we've already transcribed the audio, switching styles is instant for captions and takes just a few seconds for AI edits. Best of all, re-runs are free — no credits charged.

Track your render jobs and check where they are in the pipeline. Use these endpoints to poll for completion, view compositions, and list your render history.

Instead of polling for render status, have us notify you. Register a webhook URL and we'll send an HTTP POST whenever a render starts, completes, fails, or finishes exporting. This is especially useful for background automations, CI/CD pipelines, and AI agent workflows where you don't want to sit in a polling loop.

Webhook events:

"render.started"          — job began processing
"render.completed"        — composition ready (status = "completed")
"render.failed"           — job failed
"render.export_completed" — MP4 export done, download_url available
"render.export_failed"    — export failed

Payload example:

{
  "event": "render.completed",
  "data": {
    "id": "run_abc123",
    "status": "completed",
    "flow_type": "captions",
    "style_id": "holographic",
    "created_on": "2026-03-25T12:00:00.000Z"
  }
}

Manage your API keys and check your credit balance. You can create multiple keys for different environments (development, staging, production) and revoke any key instantly.

Purchase credits programmatically via Stripe. This is useful if you're building a platform on top of elevideo and want to manage billing from your backend.

All errors return a JSON body with error (machine-readable code) and message (human-readable description). Here's what each code means and when you might see it:

API requests are limited to 100 requests/minute per user. If you exceed this limit you'll receive a 429 response. Standard rate limit headers (RateLimit-Limit, RateLimit-Remaining, RateLimit-Reset) are included in every response.

For error handling details, see Error Codes.