To make a video on Oakgen, open the AI Video Generator, pick a model and its mode, write the prompt, set the length, shape and resolution, and press Generate. The button shows the credit price first. As of 2026-10-25 a 5-second clip costs roughly 85 credits on the cheapest setting and about 800 on Seedance 2.5 at 720p.
Every option below comes from Oakgen's live video model configs and pricing code, read on 2026-10-25. This guide also covers what happened to Cinema Studio and how to build a multi-shot video one shot at a time.
Which video models can you pick, and what do they cost?
There are seven video model families on Oakgen right now. Each runs in one or more modes, and the price is per second of output. Credits below are as of 2026-10-25 at the settings shown.
| Model | Modes | Length | Resolution | Sound | Credits per second |
|---|---|---|---|---|---|
| Seedance 2.5 | Text, image, reference | 4 to 30 s | 480p, 720p, 1080p | Generate audio toggle, on by default | 75 / 160 / 394 |
| Seedance 2.0 | Text (optional references), image | 4 to 15 s | 480p, 720p, 1080p, 4K | Toggle on image mode | 46 / 103 / 230 / 526 |
| Seedance 2.0 Fast | Text, image, reference | 4 to 15 s | 480p, 720p | Toggle, on by default | 37 / 82 |
| Kling v3 Pro | Text, image, Motion Control | 3 to 15 s | Fixed | Always on | 57 |
| Gemini Omni Flash | Text, image, reference, edit | 3 to 10 s | 720p | None | 34 |
| Grok Imagine Video | Text, image, edit, extend | 3 to 15 s | 480p, 720p | None | 17 / 24 |
| MiniMax H3 Max | Text, image, reference | 5 to 15 s | 480p, 768p, 1080p | None | 17 / 28 / 55 |
Two edit models sit in the same picker. Kling O3 Pro Video Edit changes an existing 3 to 15 second clip from a prompt at 57 credits per second. Gemini Omni Flash Edit takes a clip of up to 30 seconds at 34 credits per second. The picker also lists three utility tools: VEED Subtitles (2 credits per second), Video Clips Generator (1 credit per second) and Social Media Video Downloader.
A few pricing details matter in practice. Gemini Omni Flash holds the maximum price when you submit, then charges for the seconds it actually returns and refunds the rest. Seedance 2.5 Reference with a reference video charges 60 percent of the per-second rate, counted over the reference video's seconds plus the output seconds. Plans and add-ons are on the pricing page. As of 2026-10-30 the Basic plan is $19 a month for 4,000 credits, and plan prices can change.
Seedance 2.5 is ByteDance's newest Seedance model. ByteDance Seed announced it on 2026-07-31.
Veo, Sora, Wan, Hailuo, Runway, Luma and Vidu are not on Oakgen. If an older guide tells you to pick one of them, it is out of date.
How do you open the video generator?
Go to /ai-video-generator and sign in. The tool opens in Oakgen's Create workspace with video selected. A link to one model, such as /ai-video-generator/seedance-v2.5-text-to-video, opens the workspace with that model already chosen.
The screen has two parts. Your results fill the gallery in the middle, grouped by day or by project. The composer at the bottom holds the prompt, the model picker and the settings.
How do you choose a model and a mode?
Click the model button in the composer footer. It shows the current model's name and operation, for example "Text to video". The panel that opens is called "Choose a model". It has:
- a "Search models" box that matches names, providers and modes;
- tabs for All, Video, Image, Audio and Music;
- one row per model family, with a short description and a list price at default settings.
The mode is picked here, not with tabs on the form. A family with several modes shows an arrow. Click it to see its operations (Text to video, Image to video, Reference to video, Motion control, Video edit), each with its own price. Attaching an image to a text model does not switch it to image mode. Pick the image operation instead.
One exception: Seedance 2.0's text mode has optional reference image, video and audio slots, and Oakgen routes the job to the reference endpoint when you fill them.
Your typed prompt carries over when you switch models, so you can try the same idea on two models without retyping it.
How do you write the prompt?
Type what happens, who is in it, where it is, and how the camera moves. The box shows a character counter. Limits differ by model: 50,000 characters on Seedance and MiniMax H3 Max, 4,096 on Grok Imagine Video and 2,500 on Kling v3 Pro.
Write the camera in plain shot language. "Slow dolly in at eye level, 35mm, shallow focus" gives the model something to follow. "Cinematic" does not. Keep each prompt to one subject, one action and one place.
Two helpers sit next to the prompt:
- Type
@to pull in one of your AI Influencer characters or one of your saved assets. - The lightbulb opens "Get Inspired", which has prompt ideas and a builder that turns a rough idea into a prompt with separate camera and movement parts.
There is no one-click prompt enhancer on video. MiniMax H3 Max has its own Prompt Enhancement setting (off, fast, balanced or quality, balanced by default).
How do you set length, shape, resolution and sound?
Length, aspect ratio and the number of outputs are chips in the composer footer. Everything else, including resolution, the audio toggle, negative prompt, seed and bitrate, sits behind the sliders icon labelled "Settings".
Pick the aspect ratio before you generate. Most text modes offer 16:9 and 9:16, and Seedance also offers 21:9, 4:3, 1:1 and 3:4. Grok adds 3:2 and 2:3. Most image modes have no aspect ratio setting, so crop the start image to the shape you want. Cropping a 16:9 clip to 9:16 later throws away most of the frame.
Resolution changes the price more than anything else. Seedance 2.5 at 1080p costs about 2.5 times its 720p rate. Draft at 480p or 720p, then render the keeper at full size.
Sound depends on the model. Seedance 2.5, Seedance 2.0 Fast and Seedance 2.0 image mode have a "Generate audio" toggle that starts on. Kling v3 Pro always generates sound. Gemini Omni Flash, Grok Imagine Video and MiniMax H3 Max have no audio toggle. Turn the toggle off if you will add your own music.
How do you animate an image or use references?
Choose an image operation to start from a picture. Seedance, Kling v3 Pro and MiniMax H3 Max image modes also take an end image, so you can set the first and last frame. On Seedance 2.5 the end image is under Settings.
Choose a reference operation when you want the model to follow several inputs at once. Limits differ:
- Seedance 2.5 Reference: up to 30 images, 10 videos and 10 audio files, 50 files in total, with at least one image or video.
- Seedance 2.0 Fast Reference and MiniMax H3 Max Reference: up to 9 images, 3 videos and 3 audio files.
- Gemini Omni Flash Reference: up to 6 images.
Name the uploads in the prompt in order, as @Image1, @Video1 and @Audio1. Kling v3 Pro image mode uses "elements" instead. Each element is a front image plus up to three angle images, and a button inserts @Element1 into the prompt.
Kling v3 Pro Motion Control is different. You give it a character image and a motion video, and it copies the motion onto your character.
What happens after you click Generate?
The Generate button shows a coin and the credit price for your exact settings, and it updates as you change them. When you submit, a message confirms how many credits were charged and the job appears in the gallery. Status updates arrive live, so you can leave the page. If a job fails, the card says "Credits refunded" and offers "Try again".
Open a finished video to see its prompt, model, length and shape. From there you can:
- reuse the prompt and settings;
- use the clip as a reference for the next generation;
- download it;
- send it to the Video Upscaler, which goes up to 4K;
- open it in Compose, Oakgen's timeline editor, with "Edit video" or "Add to Compose".
How do you extend or edit a finished clip?
There is no Extend button. Use "Use as reference" on the clip, then pick a model that accepts video. Grok Imagine Extend Video adds 2 to 10 seconds to a clip of 2 to 15 seconds at 24 credits per second. Kling O3 Pro Video Edit and Gemini Omni Flash Edit change a clip from a prompt. Seedance 2.5 Reference can take the clip as @Video1 and build a new shot from it.
None of these is a region mask. They change the clip from your prompt, so describe the change and what must stay the same.
What happened to Cinema Studio, and where are its features now?
Cinema Studio is retired. It was a separate Oakgen screen where you chose a camera body, lens, focal length, aperture and a camera move, and it wrote those choices into the shot prompt for the video model. Oakgen took it out of the tool list, and /cinema-studio now permanently redirects to the AI Video Generator. The old in-app address opens the Create workspace with video selected.
Here is where each part went:
| Cinema Studio part | Where it is now |
|---|---|
| Video models (Seedance, Kling V3 Pro) | Same models in the AI Video Generator picker |
| Aspect ratio, length, resolution, reference image | Composer chips and Settings, per model |
| Camera body, lens, focal length, aperture pickers | Gone. Write them into the prompt |
| Camera-move presets (dolly in, crane up, orbit) | Gone. Write the move into the prompt, or copy motion from a video with Kling v3 Pro Motion Control |
| Voiceover and music | Separate tools: the voice generator and the music generator |
| Joining shots | Compose, or any outside editor |
The video generator has no camera presets, no color-grade presets and no ProRes export. If you relied on those, use a lens-and-move phrase in the prompt and do the grade in your editor. The 12 types of videos guide has format ideas from the Cinema Studio days that still work shot by shot.
Open the AI Video Generator
Pick a model, see the credit price on the Generate button, and render your first shot.
How do you make a multi-shot video, shot by shot?
Plan the whole piece, generate each shot as its own clip, then join them in an editor. This was the Cinema Studio method, and it works the same way in the video generator.
- Write a one-paragraph brief. State the goal (sell, explain, entertain), the tone, the runtime (30, 60 or 90 seconds) and the delivery shape (9:16 for Reels and TikTok, 16:9 for YouTube, 21:9 where Seedance offers it).
- Write the script. For ads, write voiceover lines word for word. For explainers, write beats. For story pieces, write scenes with place, subject, action and mood.
- Make a shot list with one idea per shot. Split any shot that tries to say two things. As a rough guide, a 30-second trailer has 8 to 12 shots and a 60-second ad has 10 to 16.
- Generate each shot. Choose the model, shape, length, resolution and any start image or reference. Test a key shot on two models, for example Seedance 2.0 and Kling v3 Pro, and keep the one that holds the motion, framing and subject. Check the price on the button before the final render.
- Make the audio. Generate voiceover in the voice generator and music in the music generator, then mix them in your editor.
- Assemble and export in Compose or an editor such as CapCut or DaVinci Resolve.
Here is a worked plan for a 30-second thriller teaser. The brief: "Night, rain, neon. One protagonist. No dialogue. Music-led. End on a title card."
| Shot | Description | Model and length |
|---|---|---|
| 1 | Wide rainy neon street, slow dolly in | Seedance 2.5, 5 s |
| 2 | Close-up of an eye reflecting neon, static | Kling v3 Pro, 3 s |
| 3 | Protagonist walking, tracking shot | Seedance 2.0, 5 s |
| 4 | Insert of a hand clenching | Kling v3 Pro, 3 s |
| 5 | Whip pan to a ringing phone | Kling v3 Pro, 3 s |
| 6 | Face lit by the phone screen, static | Seedance 2.0, 4 s |
| 7 | Running through an alley, handheld | Kling v3 Pro, 5 s |
| 8 | One second of black, then the title card | Added in the editor |
For audio, ask the music generator for a tense minor-key score that rises at about 60 BPM. Add rain, neon hum, phone vibration and footsteps as sound effects. Turn off the Seedance audio toggle on these shots, since the score carries the piece.
What mistakes cost the most credits?
- Over-describing a shot. Three sentences of styling for one shot gives the model too much to follow. Keep one subject, one action and one place, plus a short camera phrase.
- Choosing the shape after generating. Set 9:16 or 16:9 before the first render.
- Rendering every draft at full resolution. Draft at 480p or 720p and render the approved shot at 1080p.
- Too many cuts. For normal pacing, aim for one shot every 3 to 5 seconds. Save one-second cuts for montages.
- Drifting characters. For a recurring character, generate the first shot, then use a frame of it as the start image or a reference for later shots. Seedance 2.5 image and reference modes and Kling v3 Pro elements are built for this.
- Music over the voiceover. Duck the music under the voice in your editor. If the voice is hard to hear on laptop speakers, fix the mix.
To see how often generations fail and which models cost the most in practice, check the Oakgen AI Video Index 2026.
When is another tool a better choice?
Use something else in these cases:
- You need Veo. Every Veo model is retired from Oakgen. Google still offers Veo 3.1 on Vertex AI, so go there directly.
- You need a talking head with lip sync. Use Talking Photo, which turns a photo and a script into a talking video.
- You want a product ad with an AI presenter. UGC Ads is built for talking-avatar product ads and testimonials.
- You want help planning the shots. Agent Chat can write the shot list and call Oakgen's video models for you.
- You need exact frame control, region masks or color grading. Do that in a traditional editor. No Oakgen video model offers a region mask.
How we checked this
We read Oakgen's video model configs, the Create workspace code, the route redirects and the pricing code on 2026-10-25. Credit prices come from the live pricing code generated on 2026-10-24 and can change. The Generate button shows the current price. The Seedance 2.5 release date comes from ByteDance Seed's announcement of 2026-07-31. We did not run these models side by side for this article, and we did not time renders.
Frequently asked questions
How much does an AI video cost on Oakgen?
It depends on the model, the resolution and the length. As of 2026-10-25, a 5-second clip costs roughly 85 credits on Grok Imagine Video at 480p, about 285 on Kling v3 Pro and about 800 on Seedance 2.5 at 720p. Seedance 2.5 at 1080p is 394 credits per second. The Generate button shows the exact credit price for your settings before you run.
Which video model should I start with?
Use Seedance 2.5 for clips up to 30 seconds or a large reference pack. For cheap drafts, Grok Imagine Video at 480p costs 17 credits per second as of 2026-10-25. Kling v3 Pro is the choice when you want to copy motion from a video (Motion Control), and Seedance 2.0 is the only option with a 4K setting.
How long can one AI video be?
The longest single clip is 30 seconds, on Seedance 2.5. Most other models stop at 10 or 15 seconds. Grok Imagine Extend Video adds 2 to 10 seconds to a source clip of 2 to 15 seconds. For anything longer, generate separate shots and join them in an editor.
Is there a free plan or free credits?
No. Oakgen has no free plan and gives no credits on signup. You need a subscription or a credit add-on, both listed on the pricing page. Failed generations are refunded automatically.
Can I use the videos commercially?
Oakgen does not claim ownership of eligible outputs, but your right to use a video depends on Oakgen's Terms (section 4.4), the law, the model provider's terms and the rights in anything you upload, such as a reference photo of a real person.




