hi

Thanks for reaching out, adit-kulkarni. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

in livekit what is the difference between inference and plugin

Thanks — great question. Here’s a concise comparison between LiveKit Inference and plugins, and when to use each. What they are • LiveKit Inference: A managed LiveKit Cloud service that gives you direct access to curated STT, LLM, and TTS models via your LiveKit API key. • Plugins: Open-source integrations you install and configure that connect directly to third‑party model providers using your own accounts and API keys. Key differences • Setup: Inference works out-of-the-box with LiveKit Cloud; plugins require installing packages and supplying provider keys. • Account & billing: Inference is billed through LiveKit (no separate provider accounts). Plugins are billed and rate‑limited by the provider you connect. • Model selection & control: Inference exposes a curated set of models for simplicity and reliability. Plugins give broader provider choice and deeper features (e.g., Realtime APIs, provider-specific options). • Latency & reliability: Inference uses global co‑location and provisioned capacity for lower latency and built‑in failover. Plugin performance depends on the provider and your configuration. • Concurrency/quotas: Inference quotas are managed centrally in LiveKit; plugins rely on each provider’s quotas. When to choose which • Use LiveKit Inference if you want simple setup, unified billing, and managed reliability. • Use plugins if you need specific provider features, full control of API keys/billing, or access to providers not in the Inference catalog. If you want, tell me which models or providers you’re considering and whether you prefer LiveKit-managed billing or direct provider control — I can recommend the best path and next steps. Sources: Adrian Cowham - LiveKit Blog | Models | LiveKit Docs | Introducing LiveKit Inference: A unified model interface for voice AI | Large language models (LLM) | LiveKit Docs | Text-to-speech (TTS) models | LiveKit Docs