Updated 15 October 2026 · prices and limits reflect the current service settings
Clips from a description, a photo or references, editing your own video, plus translation, subtitles and enhancement of finished clips. 12 models; the price is per second and shown before you start.
1. Where to find it
In Telegram (Mini App)Create → Regular → AI Video; translation, subtitles and Enhance a video are in the Tools tab
On griffonhub.ruAI video menu: Text → video, Photo → video, From references, Video → video, Video translation, Subtitles, Enhance a video
2. Modes
- Text → video — a scene from a description.
- Photo → video — animate a frame. Many models also take a last frame — the clip ends on it.
- From references — characters and objects from your photos in a new scene; some models also take video and voice.
- Video → video — with Kling O3, reshoot a clip from a sample or change its style and details; with Kling 4.0, edit a finished clip with a text command.
3. Models
Seedance 2.5
flagship: clips up to 30 s, cinematic quality, sound; up to 30 reference photos
4–30 sKling 4.0
clips up to 30 s with stereo sound, 21:9, references — images, characters and video, editing your own video
3–30 sKling O3
all in one: text, photo with first and last frame, references, editing your own video
3–15 sSeedance 2.0
cinematic clips up to 4K, with sound
4–15 s, “auto”Kling v3
text and photo → video, with a 4K tier
3–15 sVidu Q4
photo → video and clips from references: up to 12 photos and 3 voices, up to 16 s and 4K
3–16 sGrok Imagine 1.5
Grok Imagine 1.5: clips from 1 second, up to 1080p
1–15 sSeedance 2.0 Fast
the fast version of Seedance 2.0, with sound
4–15 s, “auto”Grok Imagine
short clips from 1 second, sound always on
1–15 sMiniMax H3
a clip with characters and objects from your photos (up to 9)
5–15 sMiniMax H3 Turbo
inexpensive video from text or a photo
5–15 sGrok Imagine 1.5 Lite
the most affordable video: from 1 second, up to 1080p
1–15 s4. Prices and capabilities
Rates are Griffs per second of video: price = rate × duration.
| Model | Modes | Duration | Sound |
|---|---|---|---|
| Seedance 2.5 | Text → video, Photo → video, From references | 4–30 s | optional |
| Kling 4.0 | Text → video, Photo → video, From references, Video → video: edit | 3–30 s | always on |
| Kling O3 | Text → video, Photo → video, From references, Video → video: from a sample, Video → video: edit | 3–15 s | optional; video → video keeps the source sound |
| Seedance 2.0 | Text → video, Photo → video | 4–15 s, “auto” | optional |
| Kling v3 | Text → video, Photo → video | 3–15 s | optional |
| Vidu Q4 | Photo → video, From references | 3–16 s | photo → video — always; references — optional |
| Grok Imagine 1.5 | Text → video, Photo → video | 1–15 s | always on |
| Seedance 2.0 Fast | Text → video, Photo → video | 4–15 s, “auto” | optional |
| Grok Imagine | Text → video, Photo → video | 1–15 s | always on |
| MiniMax H3 | From references | 5–15 s | always on |
| MiniMax H3 Turbo | Text → video, Photo → video | 5–15 s | always on |
| Grok Imagine 1.5 Lite | Text → video, Photo → video | 1–15 s | always on |
| Model | G per second | Plan from |
|---|---|---|
| Seedance 2.5 | 480p — 40; 720p — 85; 1080p — 210 | Plus (1080p — Pro) |
| Kling 4.0 | 720p — 29; 1080p — 43 | Plus |
| Kling O3 | standard (1080p) — 15, with sound 20, video → video 23; pro (1080p) — 20, with sound 25, video → video 30 | Plus |
| Seedance 2.0 | 480p — 24; 720p — 55; 1080p — 123; 4K — 280 | Plus (1080p, 4K — Pro) |
| Kling v3 | standard (1080p) — 15, with sound 23; pro (1080p) — 20, with sound 30; 4K — 76 | Plus (4K — Pro) |
| Vidu Q4 | 540p — 8; 720p — 17; 1080p — 22; 2K — 34; 4K — 70 | Plus (1080p, 2K, 4K — Pro) |
| Grok Imagine 1.5 | 480p — 22; 720p — 38; 1080p — 68 | Plus (1080p — Pro) |
| Seedance 2.0 Fast | 480p — 21; 720p — 44 | Plus |
| Grok Imagine | 480p — 13; 720p — 19 | Plus |
| MiniMax H3 | 480p — 9; 768p — 11; 2K — 23 | Plus (2K — Pro) |
| MiniMax H3 Turbo | 480p — 5; 768p — 7 | Plus |
| Grok Imagine 1.5 Lite | 480p — 4; 720p — 6; 1080p — 26; +2 G per photo | Plus (1080p — Pro) |
- Auto duration on Seedance 2.0, Seedance 2.0 Fast is charged as 15 s.
- Seedance 2.5 with video references: lower rate (480p — 24, 720p — 51, 1080p — 126), but the length of the attached clips is added to the billed duration.
- Kling 4.0 with video input (reference videos and edits): lower rate (720p — 17, 1080p — 26); the input length plus the clip length is billed, but no less than twice the clip length; in an edit the source counts as input too.
- MiniMax H3: the first 5 photos are free, each extra one is +14 G.
- Example: 5 seconds of Grok Imagine 1.5 Lite at 720p — 30 G, Kling O3 standard with sound — 100 G.
5. Photos, video and voice as input
- Photos — up to 10 MB (Seedance 2.5 — up to 30 MB); you can pick them from the library or your works.
- Video for Kling O3 — 3–15 s, 24–60 fps, up to 95 MB.
- Video for Kling 4.0 — 3–30 s and at least 700×700 pixels — a 720p or larger clip works.
- You can refer to references in the prompt: @Image1, @Audio1 — as the tiles are labelled.
- Every model needs a prompt, except Kling O3 and Vidu Q4 in Photo → video.
- In Photo → video the frame follows the photo; on models with a format choice — with Auto.
6. Working with a finished clip
Upload a clip (up to 95 MB) or pick one of your works.
Video translation
The same voice in another language, the lips follow the new words. 39 languages, up to 5 speakers, clips of 3 s – 5 min.
Fast9 Gper second of video
Precise18 Gper second of video
A translation can run longer than the original, so it is charged with a ×1.3 reserve; the difference is refunded based on the finished translation's length.
Subtitles
Speech is recognised automatically and becomes on-screen captions: 30 styles, including animated words, at the bottom, centre or top; caption translation into 41 languages. Clips up to 10 min.
Subtitles18 Gper minute, one minute minimum, then per second
Animated styles and clips above 1080p×2
Caption translation+36 Gper minute
Enhance a video
Sharper up to 1080p4 G per secondclips up to 60 s
Sharper up to 4K11 G per secondup to 30 s, Pro and up
Sound for the picture2 G per secondup to 60 s; speech can be kept
Clean speech without noise2.25 G per minuteat least 3 G, clips up to 10 min
Background replacement4 G per 30 framesgreen screen, white, black, blue; up to 60 s
Translation, subtitles and Enhance a video are available from Plus.
7. Styles and prompts
A good prompt structure
Subject → action → setting → style → camera → mood.
“A girl in a red dress dances in the rain on a rooftop at night, neon city lights, the camera slowly circles her, cinematic”
- For photos, describe the motion — “slowly turns her head”, “wind in the hair”, “the camera pushes in”.
- Eva on the site picks a model for the task and shows the price.
8. Plans
- Free — no video.
- Plus — all models up to 768p; Kling up to 1080p.
- Pro and Max — everything, including 1080p, 2K and 4K.
9. Payment, errors and support
- The price is on the start button before anything is charged. Griffs are charged when you start, and if a generation fails they are refunded automatically.
- Finished works are in My works on the site and Works in the Mini App. Files are kept for 30–60 days — download what you want to keep.
- Parallel generations depend on the plan: Free — 1, Plus — 2, Pro — 3, Max — 4.
- Adult content, violence and anything illegal is rejected automatically.