HeyGen Voice Turbo tops the TTS leaderboard, is free on the API until Nov 30, and now has ComfyUI nodes

HeyGen announced Voice Turbo on Oct 21, and its post passed 1,800 likes within a day. Turbo is the fast mode of the in-house voice model behind HeyGen's avatar videos: the same voices with a shorter generation step. Artificial Analysis' Controlled Voice TTS Arena puts it at #1 with 1,202 Elo, ahead of HeyGen Voice at #2 and Eleven v4 Turbo. That arena has every model read the same eight cloned voices, so it compares the models, not the voice casting.

For video people the useful part is the price. HeyGen's blog says the Turbo endpoint is free through Nov 30, 2026, with no credit card. It is best effort: when free capacity runs out you get a 429 and retry later. It works with HeyGen's public voices and with instant voices cloned from one recording. Professional voice clones still need the paid heygen-voice-1 model. For production you change only the model name.

ComfyUI added three pieces on Oct 21. An Instant Clone node is free during HeyGen's preview. A Text to Speech node lets you pick Turbo or Standard. A 'HeyGen Voice 1' template clones a voice and speaks a script in one graph. ComfyUI says Standard is 50% off through Oct 31. These are cloud Partner Nodes, so they bill through Comfy rather than running locally.

Where it fits: voiceover for faceless explainers, scratch narration while you cut, and dubbing your own videos. Clone only your own voice or a voice you have written permission to use. We haven't run our own listening test yet. The leaderboard is a blind preference vote, not a measure of timing control or how well it handles long scripts, so try it on your real script before you switch a pipeline.