All articles

ElevenLabs, Flux & Suno Through One API | Varosity

Call ElevenLabs voice, Flux image, and Suno music models through a single Varosity API key. No per-provider auth. REST, MCP, and CLI supported.

One API Key for ElevenLabs, Flux, and Suno

Varosity gives developers a single API key that routes requests to ElevenLabs (voice), Flux (images), and Suno (music) without managing separate provider accounts or credentials. You authenticate once with Varosity, specify the model in your request body, and the platform handles provider-side auth, rate-limit tracking, and billing — either through your own provider keys (BYOK) or Varosity credits.

Why This Matters for Multi-Modal Pipelines

Building a pipeline that generates a product video — voiceover, background music, and visual assets — normally requires three separate SDKs, three API keys, three billing relationships, and three sets of error-handling logic. With Varosity, all three calls go to https://api.varosity.ai/v1/generate with a model field pointing at elevenlabs/eleven_multilingual_v2, flux/flux-1.1-pro, or suno/chirp-v3-5.

This matters most for:

  • AI agents that need to call media models dynamically without hardcoded provider switching
  • Backend services that produce mixed media (narrated explainer videos, audio-visual ads, game assets)
  • Rapid prototyping where swapping voice or music providers mid-project shouldn't require refactoring auth

REST Example: Voice + Image in Two Calls

# Generate voiceover with ElevenLabs
curl https://api.varosity.ai/v1/generate \
  -H "Authorization: Bearer $VAROSITY_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"elevenlabs/eleven_multilingual_v2","input":"Welcome to our product demo.","voice_id":"Rachel"}'

Generate a matching image with Flux curl https://api.varosity.ai/v1/generate \ -H "Authorization: Bearer $VAROSITY_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"flux/flux-1.1-pro","prompt":"Minimalist product shot, white background, soft lighting","aspect_ratio":"16:9"}' ```

Both calls use the same key, the same base URL, and the same response envelope. Output URLs, metadata, and credit costs return in a consistent JSON schema regardless of which underlying provider handled the request.

MCP Support for Agent-Native Workflows

If you're building with Claude, GPT-4o, or any MCP-compatible agent framework, Varosity exposes a Model Context Protocol server. Your agent can call varosity.generate as a tool, passing model name and parameters — no custom tool wrappers per provider.

{
  "tool": "varosity.generate",
  "parameters": {
    "model": "suno/chirp-v3-5",
    "prompt": "Upbeat corporate background music, 30 seconds, no vocals",
    "duration": 30
  }
}

This lets an agent orchestrate a full media production loop — write copy, generate voice, generate music, generate visuals — without leaving the tool-call interface or managing provider tokens at the agent level.

BYOK vs. Credits

Varosity supports two billing modes with no free tier:

  • BYOK (Bring Your Own Key): Attach your existing ElevenLabs, Flux, or Suno API keys in the Varosity dashboard. Varosity routes your requests using those keys; you pay providers directly at your existing rate. Varosity charges a small per-request routing fee.
  • Credits: Purchase Varosity credits and the platform covers provider costs. Useful when you don't have existing provider relationships or want consolidated invoicing.

Both modes work identically at the API level — no code changes needed to switch.

CLI Access

The Varosity CLI lets you test media generation without writing a script:

varosity generate --model elevenlabs/eleven_multilingual_v2 --input "Hello world" --voice Rachel --out ./audio.mp3

Useful for CI pipelines, local prototyping, or scripting batch asset generation.

Related Resources