Hi there! LiveKit agents in TypeScript can indeed connect with cloud-based LLMs like OpenAI and Google's Gemini through dedicated plugins, which is pretty handy. While the framework does support custom implementations of the LLM node—meaning it might be possible to work with local LLMs—there isn’t clear documentation or examples confirming this just yet. If you’re thinking about using a specific local LLM, you might want to explore some custom development within the agents-js repository to make that integration happen. Please feel free to reach out if you'd like me to walk you through the process or provide more detailed guidance—I’m here to help! Sources: agents/CHANGELOG.md | Large language model (LLM) integrations | LiveKit Docs

I'm happy to share that LiveKit offers a wonderful free tier through its Build plan on LiveKit Cloud. This plan includes monthly free quotas like these: • 1,000 agent session minutes • 1,000 SIP participant minutes • 5,000 WebRTC participant minutes • 50 GB of downstream data transfer • 60 minutes of transcode usage • 60 track egress minutes These free quotas reset every month, and the best part is, you don't need to provide a credit card to get started. Just a heads-up: if you go beyond these free limits, requests may not go through unless you decide to upgrade to a paid plan. Also, if you're interested, LiveKit's open-source core is available for self-hosting completely free of charge. Feel free to reach out if you'd like me to walk you through the details or help you get set up!