Hedra notes · Read 2026-11-06 · No lip sync
Hedra KI reads as a Hedra search, not a model name
No Hedra page read on 2026-11-06 establishes KI as a Hedra model or a language. The query sits next to Hedra, whose talking-avatar product is Hedra Avatar. This page explains that product and does not run it.
This site does not lip-sync a portrait.
Kling 2.6 is a new silent clip, not Hedra Avatar
Nano AI does not run Hedra, does not accept speech audio for lip sync, and does not make a 10-minute avatar. Kling 2.6 costs 55 credits for one new 5-second clip, with sound: false. Text and image to video here are not lip sync. Eligible new accounts receive 64 credits. A 10-second clip is 110 credits.
- Made by
- Hedra
- Run on Nano AI
- Kling 2.6 · from 55 credits per video
- Page updated
- November 6, 2026
Source: hedra.com
Hedra's own stills
Three public Hedra images saved on 2026-11-06: a use-case card, a model card, and one talking-avatar still. They are not Nano AI clips and not lip sync from this site.
Hedra KI, written as a Hedra note
The words Hedra KI do not match a model id in the API reference read on 2026-11-06. That reference calls the long-form avatar hedra-avatar. The uses page describes a talking performance from a portrait and audio, multilingual, up to ten minutes for Hedra Avatar. This page repeats those statements and stops there.
A static portrait is a different task. The AI Avatar Generator from Text page uses Nano Banana 2.1 for a still image and says it has no voice and no lip sync. Linking it is only that distinction. It is not a Hedra backend.
The box at the top opens Kling 2.6. That model can start from text or from one image. Neither path accepts speech audio, and the request sets sound: false. The pictures are Hedra's use-case card, their model card, and one published portrait still.
Hedra KI at a glance
- Query
- Hedra-related; KI not a documented model
- Their model id
- hedra-avatar
- Audio
- 0.5 to 600 seconds, up to 4 files
- Start image
- At most 10.4 MB
- Rates they print
- 2.5, 5, and 6.25 cents a second
- Kling 2.6 here
- 5 seconds 55 credits, sound: false
- Lip sync here
- Not available
What Hedra publishes about Hedra Avatar
Copied from the talking-avatar page, the model page, and the v3 API reference as read on 2026-11-06. This site did not submit a job.
- 1
Do not promote KI to a product
The official model id in the API reference is hedra-avatar. The uses page says Hedra Avatar and Hedra Character 3 are Hedra's. Nothing read that day defines a model, mode, or language called KI. KI is also a German word for AI, which is one reason the query can appear. That is a caution, not a finding that Hedra ships a German model by that name.
- 2
Portrait plus audio is their input
The uses page says you give a portrait a script and a voice, in any language, for as long as the audio runs. The API requires a prompt, aspect ratio, resolution, a start image, and audio. One speaker can omit a position. Several speakers need one audio each and a point on the start frame, in the same order. This form accepts none of that audio.
- 3
Ten minutes is their cap, not ours
The uses page says Hedra Avatar runs up to ten minutes from a single audio file. The API says each audio file is from 0.5 seconds to 600 seconds, at most 104.8 MB, and from 1 to 4 files. duration_ms is optional and at most 600000. Leave it out and they say the video follows the audio. A value trims or pads with silence. The start image is at most 10.4 MB.
- 4
Jobs and rates stay on their docs
The reference says a submit returns 202 and a job id, which you poll. The model page prints 540p at 2.5 cents a second, 720p at 5 cents a second, and 1080p at 6.25 cents a second. Aspect ratios in the schema include 1:1, 4:3, 3:4, 16:9, 9:16, 9:21, and 21:9. Those cents are Hedra's. This site does not bill them.
- 5
Kling on this site is a different request
Kling 2.6 here is text to video, or one image as a first frame. The request sets sound: false. It does not take a speech file, does not move a mouth to audio, and stops at 5 or 10 seconds. Hedra's uses page also names other models on Hedra's own key. This page does not run those either.
Official Hedra pages
The first cell is the page name. Links are the same in every language. Read on 2026-11-06.
| Page | What it states | What Nano AI does |
|---|---|---|
| Hedra Avatar API | Audio-driven jobs, 600 seconds, per-second rates | Nothing. This site does not call it. |
| Talking avatar | Portrait, voice, and up to ten minutes | Nothing. This site does not lip-sync. |
| API reference | hedra-avatar inputs, including start image and audio | Nothing. This site does not submit a job. |
Hedra Avatar and the clip this site prices
Their input is image and audio
Start image at most 10.4 MB, plus 1 to 4 audio files, each 0.5 to 600 seconds and at most 104.8 MB.
Their length follows the audio
Up to 600 seconds. duration_ms, if set, is at most 600000 and can trim or pad with silence.
Their rates are per second
The model page prints 2.5 cents at 540p, 5 cents at 720p, and 6.25 cents at 1080p. Not Nano AI credits.
Kling is not that job
55 credits, 5 seconds, sound: false. No lip sync and no 10-minute file.
Hedra Avatar next to a silent clip
Hedra figures are from their pages as read on 2026-11-06. Kling figures are prices Nano AI has set. This site did not call the Hedra API.
| Hedra Avatar | Kling 2.6 here | |
|---|---|---|
| Inputs | Portrait, prompt, and speech audio | Text, or one image as a first frame |
| Length | Up to 600 seconds, following the audio | 5 or 10 seconds |
| Mouth and voice | Audio-driven, multilingual on their page | sound: false, no lip sync |
| A price they print | 2.5, 5, or 6.25 cents a second | 55 credits for 5 seconds, 110 for 10 |
| Who runs it | Hedra, as an async job | This site, as Kling 2.6 |
Where the avatar actually runs
Use Hedra for portrait-plus-audio. Use the button here only for a new silent clip that does not follow speech.
How to use this page
- 1
For a talking portrait, open Hedra
Use their talking-avatar page or the hedra-avatar API. Bring a portrait and audio there. This site will not do that step.
- 2
For a new silent clip, open Kling 2.6
The button selects 5 seconds. The price is 55 credits and the request sets sound: false. An image would be a first frame, not a face driven by speech.
- 3
Do not read KI as a language pack
Until Hedra documents that name, treat the query as Hedra plus an extra token. The German word KI is a possible reason, not a product fact.
Prompts for a new silent clip
Scenes for Kling 2.6. They do not ask for lip sync, a voice, or a ten-minute avatar.
Empty chair
An empty chair by a bright window, plant leaves moving slightly, one continuous shot, no people, no readable text, no logos.
Desk mic
A microphone on an empty desk, dust in a shaft of light, camera still, no people, no readable text, no logos.
City dusk
A quiet side street at dusk, one shop light turning on, no people, no readable text, no logos.
First frame
The camera holds on the object in the first frame and drifts slightly closer. No speech, no captions, no logos, no people.
What this page will not do
Hedra's limits are theirs. Hitting them here is not possible because the request is never sent.
- No speech upload
- Kling 2.6 on this site does not take an audio file. A picture is a first frame for a new silent clip, not a mouth to drive.
- No ten-minute avatar
- The priced lengths are 5 seconds for 55 credits and 10 seconds for 110. 600 seconds is a Hedra API limit, not a control on this button.
- KI is not filled in
- If a later Hedra page defines KI as a model or a language, this note should be revised. The 2026-11-06 reading did not find that page.
- A still is not a result
- The portrait below is Hedra's published image. It does not move, and this site did not generate it.
This site does not lip-sync a portrait.
Kling 2.6 is a new silent clip, not Hedra Avatar
Nano AI does not run Hedra, does not accept speech audio for lip sync, and does not make a 10-minute avatar. Kling 2.6 costs 55 credits for one new 5-second clip, with sound: false. Text and image to video here are not lip sync. Eligible new accounts receive 64 credits. A 10-second clip is 110 credits.
FAQ
Hedra KI FAQ
Why the query is treated as Hedra, and what this site refuses to pretend.
Is KI a Hedra model or language?
Not on the pages read on 2026-11-06. The documented id is hedra-avatar. KI can be the German word for AI, so the query looks Hedra-related. This page does not upgrade that guess into a product.
Does Nano AI run Hedra Avatar?
No. It does not accept speech audio, does not lip-sync a portrait, and does not render 10 minutes. The button opens Kling 2.6 for 55 credits and 5 seconds, sound: false.
How long is Hedra Avatar?
Their uses page says up to ten minutes from one audio file. The API says 0.5 to 600 seconds per file, up to four files, and duration_ms at most 600000. That is their limit.
What does a second cost on Hedra?
The model page, read on 2026-11-06, prints 2.5 cents at 540p, 5 cents at 720p, and 6.25 cents at 1080p. Those are not Nano AI credits.
Is the portrait below a person this site filmed?
No. It is a still Hedra published on the talking-avatar page. It is not labeled as a public figure here, and it does not move.
Is Nano AI affiliated with Hedra?
No. Hedra's API and Studio are the place for their avatar. Kling text or image to video on this site is a different, silent clip.
More from Nano AI
Related pages
Nano AI is an independent service and is not affiliated with Hedra. Avatar limits and rates are from Hedra's pages as read on 2026-11-06. The linked generator runs Kling 2.6 and does not lip-sync.