The Pulse
HeyGen Makes Its No. 1 Voice Model Free Through November 30
HeyGen says its Voice Turbo text-to-speech model is free through November 30, 2026, for developers using the company’s API. The model ranks first on Artificial Analysis’ controlled-voice leaderboard, where its score interval overlaps with H
AI.info Team · Written with AI · fact-checked ·
“We built HeyGen Voice to deliver the best possible AI voice that technology can deliver; one that always sounds like you at your best.”
Rong Yan, CTO of HeyGen, in the company’s Oct. 9 launch announcement.
Free access has a November deadline
HeyGen made its Voice Turbo text-to-speech model free through its API on Oct. 21, with the offer scheduled to run through Nov. 30, 2026. The company says the preview costs $0 per million characters and requires no credit card. Developers need a HeyGen account and API key, then can select the model ID heygen-voice-1-turbo-free in a text-to-speech request. HeyGen says the request returns a link to a WAV file, so developers can test the service with a single API call.
Turbo is the faster mode of HeyGen Voice, the company’s in-house speech model. HeyGen describes it as using the same voices with a shorter generation step. The free version supports HeyGen’s public voices and instant voice clones, which can be created from a recording during the preview. Professional voice clones are excluded; HeyGen directs users to its standard heygen-voice-1 model for those.
What the No. 1 ranking measures
HeyGen says Voice Turbo topped Artificial Analysis’ controlled-voice text-to-speech leaderboard in a snapshot captured Oct. 20, with HeyGen Voice in second place. The comparison uses the same eight cloned voices for each model: four US English and four UK English. HeyGen describes the results as blind listening tests in which people compare two audio clips and choose the one they prefer.
Artificial Analysis’ leaderboard lists Turbo at 1,202 Elo, with a reported 95% interval of 1,183 to 1,221, from 1,685 samples. HeyGen Voice scores 1,195, with an interval of 1,180 to 1,210, from 1,964 samples. Both entries appear within the leaderboard’s rank range of first to third, and their score intervals overlap, so the table’s first-place ordering does not show a clear separation between the two HeyGen models. The leaderboard is a controlled comparison of cloned English voices, not a measure of every voice, language or use case.
Free does not mean production-guaranteed
HeyGen positions the free tier for testing and prototypes, not guaranteed production service. Capacity is shared and limited, and the company offers no uptime or latency guarantee. When capacity fills, requests can receive an HTTP 429 response; HeyGen advises users to retry later. Its comparison table lists the paid Turbo model at $4 per million characters, with priority speed, and standard HeyGen Voice at $15 per million characters.
The company says developers moving to its paid Turbo model can change the model name without changing the rest of the request. That means the free offer removes the per-character charge for a limited preview, but does not remove capacity limits or provide service guarantees. The free API model is set to expire on Nov. 30, 2026.