![]()
The HeyGen Avatar 6 API on PoYo has two model IDs: heygen-avatar-6, HeyGen's Avatar 6 Quality tier, and heygen-avatar-6-turbo. They share one endpoint, one request format, and every limit, so the choice is a per-request routing decision about cost and quality, not an integration project. This page sets out what is the same, what differs, what we measured, and a rule for picking. Both are on the HeyGen Avatar 6 model page.
What is identical in the two HeyGen Avatar 6 API tiers
PoYo's HeyGen Avatar 6 docs describe Turbo as the "Turbo tier of Avatar 6 with the same inputs and modes." HeyGen's own developer docs say the same: both models "take the same inputs."
| Item | heygen-avatar-6 | heygen-avatar-6-turbo |
|---|---|---|
| Endpoint | POST /api/generate/submit | Same |
Photo mode (image_urls + audio_url) | Yes | Yes |
Reference video mode (video_url + audio_url) | Yes | Yes |
Public avatar mode (avatar + text/voice or audio_url) | Yes | Yes |
| Avatars and voices | 1,248 looks, 102 voices | Same list |
| Audio | 2–60 s, up to 15 MB | Same |
| Text | Up to 1,000 characters | Same |
prompt | Up to 1,200 characters | Same |
resolution | 1080p (default), 2K | Same |
aspect_ratio (public avatar) | 16:9, 9:16, 1:1, 4:5, 5:4 | Same |
| Billing rules | Rounded-up seconds; audio charged at submission, text reserved then settled | Same |
Because nothing in input changes, a request body that works on one model works on the other. The switch is the model string.
What differs: price
1 credit = $0.005, per output second:
| Resolution | heygen-avatar-6 | heygen-avatar-6-turbo | Turbo saves |
|---|---|---|---|
| 1080p | 20 credits ($0.10) | 10 credits ($0.05) | 50% |
| 2K | 29 credits ($0.145) | 18 credits ($0.09) | 38% |
Price per same clip, with only the model ID changed:
| Clip | heygen-avatar-6 | heygen-avatar-6-turbo |
|---|---|---|
| 10 s, 1080p | 200 credits ($1.00) | 100 credits ($0.50) |
| 10 s, 2K | 290 credits ($1.45) | 180 credits ($0.90) |
| 30 s, 1080p | 600 credits ($3.00) | 300 credits ($1.50) |
| 60 s, 1080p | 1,200 credits ($6.00) | 600 credits ($3.00) |
| 60 s, 2K | 1,740 credits ($8.70) | 1,080 credits ($5.40) |
One more comparison that affects routing: Turbo at 2K (18 credits per second) is cheaper than Avatar 6 at 1080p (20). If you need the pixel count more than the Quality tier, Turbo 2K is the cheaper way to get it.
Details in HeyGen Avatar 6 API pricing.
What differs: quality and speed, by HeyGen's account
HeyGen names the tiers "Avatar 6 Quality" and "Avatar 6 Turbo" in its technical report but does not define the difference between them, and it publishes no latency figures for Turbo. What it does publish is each tier against its own predecessor: 69.7% of 879 votes preferred Avatar 6 Quality over Avatar V, and 64.3% of 863 votes preferred Avatar 6 Turbo over Avatar V Turbo. HeyGen did not publish a Quality-vs-Turbo vote, so there is no official number for how much you give up with Turbo.
What we measured on launch day
The PoYo team ran one text request on each model on the production API on November 19, 2026. Wall time is submit to result for a client polling every few seconds, so treat it as approximate.
heygen-avatar-6-turbo | heygen-avatar-6 | |
|---|---|---|
| Avatar / voice | Abigail Office Front / Jessica Anne Bogart | Bryan Tech Expert / Andrew |
| Script | 27 words | 21 words |
| Settings | 16:9, 1080p | 16:9, 1080p |
| Output length | 10.16 s | 6.88 s |
| Cost at current prices | 110 credits (11 s × 10) | 140 credits (7 s × 20) |
| Wall time | About 2.4 min | About 2 min |
These are two single runs with different scripts and avatars, so they do not show Turbo being faster or slower. They do show the cost gap in practice: the longer Turbo clip still cost less than the shorter Avatar 6 clip. They also show that both tiers take minutes, so neither is a synchronous, wait-on-the-request API. Plan for polling or a callback_url either way.
Which HeyGen Avatar 6 API model ID to call
| Situation | Model ID | Why |
|---|---|---|
| Drafts, script and timing reviews | heygen-avatar-6-turbo | Half the price at 1080p; same fields, so the approved body re-runs on Avatar 6 unchanged |
| High-volume personalized clips | heygen-avatar-6-turbo | Per-second cost dominates at scale |
| Users previewing in your app | heygen-avatar-6-turbo | Cheap iterations while the user tries avatars and scripts |
| Final presenter or brand video | heygen-avatar-6 | HeyGen's Quality tier |
| Need 2K on a budget | heygen-avatar-6-turbo + 2K | 18 credits per second, below Avatar 6 at 1080p |
| Final 2K master | heygen-avatar-6 + 2K | Top tier at the top resolution |
A two-step pipeline is the natural fit: render every draft on Turbo, store the exact request body, and re-submit the approved body with "model": "heygen-avatar-6". Both requests below are identical apart from the model:
{
"model": "heygen-avatar-6-turbo",
"input": {
"avatar": "Abigail Office Front",
"text": "Hello and welcome to this demonstration of HeyGen Avatar 6.",
"voice": "Warm Pro Narrator",
"aspect_ratio": "16:9",
"resolution": "1080p"
}
}
{
"model": "heygen-avatar-6",
"input": {
"avatar": "Abigail Office Front",
"text": "Hello and welcome to this demonstration of HeyGen Avatar 6.",
"voice": "Warm Pro Narrator",
"aspect_ratio": "16:9",
"resolution": "1080p"
}
}
With text input, the final render is a new generation, so its length can differ slightly from the draft. With an audio file, both renders run the length of the same audio and cost exactly what the table predicts. For the full request flow, see the HeyGen Avatar 6 API guide.
Errors and retries are the same on both tiers
Routing between the two IDs does not change error handling. Both return the same body code values: 400 for an input problem (audio outside 2–60 s, an unknown avatar name, fields from two modes), 402 insufficient_credits_error when the balance is short, 429 for rate limits, and 5xx for transient failures. A 400 on Turbo will also be a 400 on Avatar 6, so never "retry on the other tier" to fix a validation error. Fix the input instead.
Reservations apply across both tiers. Text tasks on either model hold credits from submission until they settle, and the 402 message counts all of them together. In our launch-day test, after a Turbo text task and an Avatar 6 text task had been submitted, a third text request (Turbo, 2K) was refused with HTTP 200 and body code 402: "this request needs at least 252 credits, and 344 credits are already reserved by your unfinished HeyGen Avatar 6 text tasks." Budget your balance for every text task in flight, whichever model it runs on.
A routing function
import math
def pick_model(final: bool) -> str:
"""Route a HeyGen Avatar 6 request. Fields stay identical; only the model changes."""
return "heygen-avatar-6" if final else "heygen-avatar-6-turbo"
PRICE_PER_SECOND = { # credits, 1 credit = $0.005
("heygen-avatar-6-turbo", "1080p"): 10,
("heygen-avatar-6-turbo", "2K"): 18,
("heygen-avatar-6", "1080p"): 20,
("heygen-avatar-6", "2K"): 29,
}
def estimate_credits(model: str, resolution: str, audio_seconds: float) -> float:
return math.ceil(audio_seconds) * PRICE_PER_SECOND[(model, resolution)]
estimate_credits is exact for audio requests, which are billed on the rounded-up audio length. For text requests it is only an estimate, because the billed length is the finished video.
Frequently asked questions
What is the difference between the HeyGen Avatar 6 API and HeyGen Avatar 6 Turbo?
Price and HeyGen's tier. The fields, modes, limits, avatars, and voices are the same. Turbo costs 10 credits ($0.05) per second at 1080p versus 20 ($0.10), and 18 ($0.09) at 2K versus 29 ($0.145).
Is HeyGen Avatar 6 Turbo faster?
HeyGen's technical report and developer docs publish no latency numbers for either tier. In our two launch-day runs, both took about 2 to 2.4 minutes for a short clip, with different scripts, so treat speed as something to measure on your own requests.
Can I switch from Turbo to Avatar 6 without changing my request?
Yes. Change model from heygen-avatar-6-turbo to heygen-avatar-6; everything in input stays valid.
Does HeyGen Avatar 6 Turbo support 2K?
Yes, with "resolution": "2K", at 18 credits ($0.09) per second.
Which HeyGen Avatar 6 API tier should I start with?
Start on Turbo while you tune scripts, prompts, and avatar choice, then send final renders to heygen-avatar-6. The text-to-avatar walkthrough is a cheap place to start.