TruGen AI vs Talking Avatar: Pricing, LiveKit Avatars and Limits

As of 30 October 2026, TruGen sells monthly credit plans: Free with 1,500 credits, Starter at $9 for 3,000, Pro at $36 for 15,000 and Scale at $199 for 100,000, plus Enterprise by quote with "scaling discounts". A minute of TruGen's full agent uses 50 credits, and a minute of its face on your own LiveKit agent uses 40. So Starter buys 60 end-to-end or 75 face-only minutes, Pro 300 or 375, and Scale about 2,000 or 2,500. Each plan card lists a per-minute rate: $0.15, $0.12 and $0.10 end to end, and $0.12, $0.10 and $0.08 for the face. Those equal the plan price divided by its minutes ($199 ÷ 2,500 is $0.0796); the page doesn't say outright how minutes past the plan are billed. Ignore listings that quote an "Essential" plan at $99: it isn't in TruGen's current plan list.

TruGen prices two products. "Voice to video" is the face alone: TruGen's LiveKit Agents plugin adds a participant to your room that lip-syncs the audio your agent already produces, and your speech recognition, LLM and TTS stay as they are. "End to end" is TruGen's own agent: its pipeline runs speech recognition, an LLM (Groq, Azure OpenAI, Google, OpenRouter, LiveKit Inference or your own OpenAI-compatible model), a voice from ElevenLabs, Cartesia, Rime, Fish Audio or Inworld, a knowledge base, tools and the Huma-2 avatar. That costs 25% more per minute. Two things follow. If you already run a voice agent, compare TruGen's voice-to-video rate with ours, $0.08–0.12 against $0.02. If you don't, compare its end-to-end rate with Talking Avatar Voice Agents, $0.10–0.15 against $0.06, keeping in mind that TruGen adds camera vision and meeting bots.

Assume extra minutes bill at the rate on each plan card. For a face on your own agent, 1,000 minutes a month cost $98.50 on TruGen Pro ($36 plus 625 minutes at $0.10); Starter plus extra minutes would be $120. Above about 2,000 minutes, Scale is the cheaper plan, and 10,000 minutes cost $799 ($199 plus 7,500 at $0.08). On Talking Avatar Realistic Avatars the same months cost $20 and $200. With the conversation included, TruGen end to end costs $120 for 1,000 minutes on Pro and $999 for 10,000 on Scale, against $60 and $600 on Voice Agents. TruGen's lowest published face-only rate, $0.0796, is about four times ours at any volume. At enterprise volume both sides quote: our enterprise and high-volume contracts go as low as $0.005 a minute, and TruGen Enterprise doesn't publish rates.

TruGen caps concurrency and call length by plan: 1 session at a time and 5-minute calls on Free, 2 and 20 minutes on Starter, 4 and 60 minutes on Pro, and 20 sessions with no stated length limit on Scale. Enterprise lifts the concurrency cap. The caps can pick the plan before your minutes do. A support widget that sees 300 minutes a month but must answer five visitors at once needs Scale at $199, because Pro allows four. A tutoring or interview product with 45-minute sessions can't run on Starter at all. Talking Avatar Realistic Avatars runs unlimited sessions at once, so fifty cost the same per minute as one, and each session runs up to 3 hours. On TruGen Scale, with a steady peak under 20, the limits stop mattering and the choice comes down to price per minute.

TruGen's face-only mode lives in LiveKit. In Python you install livekit-agents[trugen], create trugen.AvatarSession with an avatar ID and start it with your agent session and room; in Node.js the package is @livekit/agents-plugin-trugen. The avatar joins as a participant (identity trugen-avatar) and stops when the room disconnects. The plugin is maintained in LiveKit's agents repository (version 1.8.10, released 28 October 2026). We found no TruGen plugin for Pipecat and no other face-only path in its docs. Talking Avatar Realistic Avatars isn't tied to LiveKit. Your server pushes PCM16 audio at 8 to 48 kHz from any agent (OpenAI Realtime, Pipecat, a TTS stream), through our Node.js SDK or a WebSocket API that any language can call, and the video plays in your page over WebRTC. On LiveKit, the avatar joins your room with a publish-only token you mint, so your secrets stay with you. Python LiveKit agents use our realtime API, with a complete example in our docs, rather than a one-line plugin.

TruGen renders its stock avatars with Huma-2, a model built on Gaussian splatting, and says the render takes 500 ms or less from audio in to video out, with about 800 ms end to end on its default Groq and ElevenLabs stack. Custom avatars, made from one photo or a camera capture, need its newer Huma-3 model, which the docs call beta, and the Scale plan lists "limited custom avatars". Talking Avatar Realistic Avatars offers a growing library of ready-made photorealistic avatars, and a custom avatar of your own presenter costs $199 once, with their consent. We measure less than 250 ms from your agent's audio reaching our servers to the avatar speaking it. Both are vendor figures that leave out your network and your agent's own delay, so test both with your own agent.

If you use TruGen's LiveKit plugin, your agent already makes the audio, and switching changes only the face. If TruGen's end-to-end agent runs the conversation, move that part first, to your own stack or to Voice Agents.

1. Create a Talking Avatar Realistic Avatars account (it comes with $1 of free credits), create an API key and pick an avatar ID from the library, or order a custom avatar for $199.
2. Remove trugen.AvatarSession from your LiveKit agent and keep the AgentSession, speech recognition, LLM and TTS as they are.
3. Mint a publish-only LiveKit token for the avatar and start the session with it: through our Node.js SDK, or through the realtime API from Python, following the Python example in our docs.
4. Send your agent's TTS audio to the session as PCM16 at its native rate (8 to 48 kHz), and stop playing the agent's original voice track, since the avatar publishes the voice in sync with its lips.
5. Close each session when the conversation ends; billing runs per second until you do, for up to 3 hours a session.
6. Run both on a slice of traffic for a week and compare cost per conversation before you cancel the TruGen plan.