M
Go Live
Configure

Providers / Keys

MaskRoom orchestrates external AI and render providers. Run on simple built-in credits, or bring your own keys for direct control and billing.

Connect your AI keys

OpenAI-compatible endpoints — paste your keys and they're stored in your browser. Character replies and asset generation go live the moment a key is set.

Bring your own keys

Text (chat / persona)

Image

Voice / Render

Built-in creditsSimple

MaskRoom-managed providers. One credit balance, nothing to configure. Best to get live fast.

Bring your own keysAdvanced

Connect your own accounts and endpoints. You control rate limits, models, and billing directly with each provider.

Render Engine

External render engine

You bring the real-time face-swap / portrait engine. MaskRoom provides the config surface, health checks, and session lifecycle.

Local Render Engine
Connected·Deep-Live-Cam· 190ms
BYOK
Cloud Render Engine
Not checked
BYOK

Compatible engines include Deep-Live-Cam, DeepFaceLive, FasterLivePortrait, and Chinese digital-human engines. MaskRoom never ships the engine itself — you run it locally or in your own cloud and point us at its endpoint.

Provider interface
interface RenderEngineProvider {
  status(endpoint: string):
    Promise<{ connected: boolean; engine: string }>;
  startSession(config: {
    characterId: string;
    referenceImageUrl: string;
    mode: string;
    cameraDeviceId?: string;
  }): Promise<{ sessionId: string; estimatedCreditsPerMinute: number }>;
  stopSession(sessionId: string): Promise<void>;
}

Any engine that implements this shape can be wired in. The health check calls status(endpoint).

AI Providers

Anthropic Claude
TextBuilt-in credits
Connected

Persona rewrite + chat assist

OpenAI-compatible · Anthropic · custom endpoint

Latency 320ms
OpenAI-compatible
TextBYOK
Off

Fallback text provider

OpenAI-compatible · Anthropic · custom endpoint

Replicate / FAL
ImageBuilt-in credits
Connected

Reference pack + asset generation

Replicate · FAL · RunPod · custom endpoint

Latency 4200ms
ElevenLabs
VoiceBYOK
Connected

Voice cloning + TTS

ElevenLabs · OpenAI-compatible · custom endpoint

Latency 540ms