Generate AI video now: Open the Imagera AI Video Generator → — Kling 3.0 Turbo, Seedance 2.0 Mini, HappyHorse 1.1, and 10+ other video models in one dashboard, no GPU required.
Three new video models landed on Imagera AI in mid-2026: Kling 3.0 Turbo, Seedance 2.0 Mini, and HappyHorse 1.1. Each targets a different creative problem and a different credit budget. This guide breaks down every relevant dimension — credits per second, resolution ceiling, audio behavior, reference input types, aspect ratio support, and generation mode coverage — so you can pick the right model before you spend a single credit.
The quick answer: Kling 3.0 Turbo runs at 30 cr/s at 720p — 25% lower than the other two at the same resolution. Seedance 2.0 Mini is the only model with 480p output (20 cr/s) and multimodal reference inputs. HappyHorse 1.1 is the only model with native multilingual lip-sync audio and up to 9 reference images for character consistency across shots.
Read on for the full breakdown with per-scenario credit math.

1.At a Glance: The Three Models
| Feature | Kling 3.0 Turbo | Seedance 2.0 Mini | HappyHorse 1.1 |
|---|---|---|---|
| Provider | Kuaishou | ByteDance | Alibaba |
| Tier | Budget / Turbo | Budget / Mini | Premium |
| Resolutions | 720p · 1080p | 480p · 720p | 720p · 1080p |
| Credits/s @ 720p (Kie route) | 30 cr/s | 40 cr/s | 40 cr/s |
| Credits/s @ 1080p (Kie route) | 35 cr/s | — | 65 cr/s |
| Credits/s @ 480p (Kie route) | — | 20 cr/s | — |
| Base display credits (5 s default) | 150 cr | 200 cr† | 325 cr |
| Duration range | 3–15 s | 5 / 10 / 15 s | 5 / 10 / 15 s |
| Native audio | Yes (sound) | Yes | Yes (joint, multilingual) |
| Lip-sync audio | No | No | Yes (multilingual) |
| Generation modes | T2V · I2V | T2V · I2V · R2V | T2V · I2V · R2V |
| Max reference images | — | Multimodal† | Up to 9 |
| Aspect ratios | 3 | 6 | 9 |
† Seedance 2.0 Mini base display credits are 400 cr (40 cr/s × 10 s default). For a 5 s clip at 720p the cost is 200 cr (40 × 5). Seedance 2.0 Mini R2V accepts image, video, and audio references.
T2V = Text-to-Video. I2V = Image-to-Video. R2V = Reference-to-Video.
2.Kling 3.0 Turbo — Multi-Shot Storytelling at 30 Credits per Second
Kling 3.0 Turbo is Kuaishou's fast, lower-cost variant of Kling 3.0. Its headline capability is multi-shot storytelling: the model accepts up to six connected shots via the
multi_prompt parameter, so you can describe a sequence of scenes in a single generation pass instead of stitching individual clips together in post.
2.1Kling 3.0 Turbo Capabilities
| Dimension | Detail |
|---|---|
| Generation modes | Text-to-Video (T2V) · Image-to-Video (I2V) |
| Resolutions | 720p (Standard) · 1080p (Pro) |
| Durations | 3 s through 15 s in single-second increments |
| Aspect ratios | 16:9 · 9:16 · 1:1 |
| Native audio | Yes — audio generated alongside video |
| Multi-shot | Up to 6 connected shots via |
| Reference-to-Video | Not available — Turbo is T2V + I2V only |
2.2Kling 3.0 Turbo Pricing (Kie Route, Normal Mode)
Credits are billed per second, per resolution. The numbers below come from the verified Imagera pricing SSOT (KLING_3_TURBO_PRICING, Kie route, 2026-06-30):
| Resolution | Credits per second | Example: 5 s | Example: 10 s | Example: 15 s |
|---|---|---|---|---|
| 720p (Standard) | 30 cr/s | 150 cr | 300 cr | 450 cr |
| 1080p (Pro) | 35 cr/s | 175 cr | 350 cr | 525 cr |
At 720p, Kling 3.0 Turbo runs at 30 cr/s — 25% lower than both Seedance 2.0 Mini and HappyHorse 1.1 at the same resolution (each at 40 cr/s). At 1080p, it runs at 35 cr/s versus 65 cr/s for HappyHorse 1.1 on the Kie route — a 46% difference per second of output.
2.3When to Choose Kling 3.0 Turbo
- You are generating a sequence of narrative scenes and want multi-shot continuity in a single generation
- Credit efficiency at 720p or 1080p is the primary constraint
- You do not need reference images or multimodal input (R2V not available)
- You want the widest duration flexibility (3 s steps from 3 to 15 s)
3.Seedance 2.0 Mini — Multimodal References at 480p/720p
Seedance 2.0 Mini is ByteDance's cost-optimized tier of Seedance 2.0. Its distinguishing feature is multimodal reference-to-video (R2V): it accepts image, video, and audio reference inputs simultaneously, letting you constrain the output against visual style, motion style, and audio mood in a single call.
It is the only model of the three that outputs 480p, which makes it the lowest-credit-per-second option for storyboarding, thumbnail previews, or any use case where resolution is secondary to idea validation.
3.1Seedance 2.0 Mini Capabilities
| Dimension | Detail |
|---|---|
| Generation modes | Text-to-Video (T2V) · Image-to-Video (I2V) · Reference-to-Video (R2V) |
| Resolutions | 480p · 720p (no 1080p — this is the Mini / budget tier) |
| Durations | 5 s · 10 s · 15 s |
| Aspect ratios | 16:9 · 9:16 · 1:1 · 4:3 · 3:4 · 21:9 |
| Native audio | Yes ( flag, default on) |
| Reference inputs (R2V) | Image references · Video references · Audio references |
| Lip-sync audio | No |
3.2Seedance 2.0 Mini Pricing (Kie Route, Normal Mode)
Credits come from the verified Imagera pricing SSOT (SEEDANCE_2_0_MINI_PRICING, Kie route, 2026-06-30). The default duration is 10 seconds (base display = 400 credits), but you can generate shorter clips:
| Resolution | Credits per second | Example: 5 s | Example: 10 s | Example: 15 s |
|---|---|---|---|---|
| 480p | 20 cr/s | 100 cr | 200 cr | 300 cr |
| 720p | 40 cr/s | 200 cr | 400 cr | 600 cr |
At 480p (20 cr/s), Seedance 2.0 Mini has the lowest credit rate of the three models. At 720p it matches HappyHorse 1.1 at 40 cr/s, and costs 33% more per second than Kling 3.0 Turbo (30 cr/s).
3.3When to Choose Seedance 2.0 Mini
- You need multimodal reference inputs — video or audio references alongside image references
- You want to constrain visual style or motion from a reference clip
- You are generating storyboard previews or concept drafts where 480p is sufficient (100 cr per 5 s)
- Credit rate at 720p is not the deciding factor (equals HappyHorse 1.1 at the same resolution)
4.HappyHorse 1.1 — Native Multilingual Lip-Sync and 9-Image Reference Control
HappyHorse 1.1 is Alibaba's cinematic video model with native, synchronized joint audio that includes multilingual lip-sync in a single pass. Unlike Seedance 2.0 Mini, which has a
generate_audio on/off flag, HappyHorse 1.1 always generates audio jointly with video — the audio cannot be disabled and is not user-controllable beyond the text prompt. This makes it the right choice when synchronized dialogue or multilingual narration is the core requirement.
The R2V mode accepts up to 9 reference images, which you address in the prompt as
[Image 1] through [Image 9]. This is the highest reference image capacity of the three models, making HappyHorse 1.1 the strongest option for maintaining character or object consistency across multiple shots.
4.1HappyHorse 1.1 Capabilities
| Dimension | Detail |
|---|---|
| Generation modes | Text-to-Video (T2V) · Image-to-Video (I2V) · Reference-to-Video (R2V) |
| Resolutions | 720p · 1080p (no 480p) |
| Durations | 5 s · 10 s · 15 s |
| Aspect ratios | 16:9 · 9:16 · 1:1 · 4:3 · 3:4 · 21:9 · 9:21 · 5:4 · 4:5 (9 ratios — widest of the three) |
| Native audio | Always on — joint generation, multilingual lip-sync |
| Reference inputs (R2V) | Up to 9 image references (image only — no video/audio reference inputs) |
| Lip-sync audio | Yes — multilingual |
4.2HappyHorse 1.1 Pricing (Both Routes)
HappyHorse 1.1 pricing diverges between the Kie (Normal) and Fal (Ultra Fast) routes because the underlying wholesale costs differ. Numbers from the verified SSOT (HAPPY_HORSE_1_1_PRICING, 2026-06-30):
| Route | Resolution | Credits per second | Example: 5 s | Example: 10 s | Example: 15 s |
|---|---|---|---|---|---|
| Kie (Normal) | 720p | 40 cr/s | 200 cr | 400 cr | 600 cr |
| Kie (Normal) | 1080p | 65 cr/s | 325 cr | 650 cr | 975 cr |
| Fal (Ultra Fast) | 720p | 35 cr/s | 175 cr | 350 cr | 525 cr |
| Fal (Ultra Fast) | 1080p | 45 cr/s | 225 cr | 450 cr | 675 cr |
HappyHorse 1.1 costs the most credits per second at 1080p on the Kie route (65 cr/s vs 35 cr/s for Kling Turbo). The Fal Ultra Fast route closes that gap — 45 cr/s at 1080p vs 35 cr/s for Kling Turbo on the same route.
4.3When to Choose HappyHorse 1.1
- Synchronized lip-sync audio in any language is a requirement
- You need up to 9 character/object reference images for multi-image consistency
- You need the widest aspect ratio selection (9 options, including 21:9, 9:21, 5:4, 4:5)
- Credit efficiency is secondary to cinematic output quality and audio fidelity
- You want 1080p output with the Ultra Fast (Fal) route at 45 cr/s instead of 65 cr/s
5.Credit-Per-Second Comparison at Each Resolution
This table compares all three models on the Kie (Normal) route — the default route on Imagera — at each resolution they share or uniquely support:

| Resolution | Kling 3.0 Turbo | Seedance 2.0 Mini | HappyHorse 1.1 |
|---|---|---|---|
| 480p | — | 20 cr/s | — |
| 720p | 30 cr/s | 40 cr/s | 40 cr/s |
| 1080p | 35 cr/s | — | 65 cr/s |
Key takeaways:
- At 480p: Seedance 2.0 Mini exclusively — 20 cr/s, the lowest credit rate across all three models at any resolution
- At 720p: Kling 3.0 Turbo runs at 30 cr/s; Seedance Mini and HappyHorse 1.1 are tied at 40 cr/s (33% higher than Kling Turbo)
- At 1080p: Kling 3.0 Turbo costs 35 cr/s; HappyHorse 1.1 costs 65 cr/s on the Kie route (86% higher), or 45 cr/s on the Fal Ultra Fast route
6.Side-by-Side Cost for Common Clip Lengths
To make the credit math concrete, here are the Kie-route costs for common generation scenarios.
6.15-Second Clip
| Resolution | Kling 3.0 Turbo | Seedance 2.0 Mini | HappyHorse 1.1 |
|---|---|---|---|
| 480p | — | 100 cr | — |
| 720p | 150 cr | 200 cr | 200 cr |
| 1080p | 175 cr | — | 325 cr |
6.210-Second Clip
| Resolution | Kling 3.0 Turbo | Seedance 2.0 Mini | HappyHorse 1.1 |
|---|---|---|---|
| 480p | — | 200 cr | — |
| 720p | 300 cr | 400 cr | 400 cr |
| 1080p | 350 cr | — | 650 cr |
6.315-Second Clip
| Resolution | Kling 3.0 Turbo | Seedance 2.0 Mini | HappyHorse 1.1 |
|---|---|---|---|
| 480p | — | 300 cr | — |
| 720p | 450 cr | 600 cr | 600 cr |
| 1080p | 525 cr | — | 975 cr |
7.Capability Gaps: What Each Model Cannot Do
Understanding the hard limits prevents wasted generations:
| Limitation | Kling 3.0 Turbo | Seedance 2.0 Mini | HappyHorse 1.1 |
|---|---|---|---|
| No R2V mode | Yes — T2V + I2V only | No | No |
| No 480p output | Yes | No | Yes |
| No 1080p output | No | Yes — 720p ceiling | No |
| No lip-sync audio | Yes | Yes | No — always included |
| No user audio control | No — sound flag | No — generate_audio flag | Yes — joint audio, no flag |
| No video/audio references | Yes | No — supports all ref types | Yes — image references only |
| No 9-image reference | Yes | Yes | No — up to 9 |
8.Use-Case Decision Guide

8.1You Need Multi-Shot Narrative Video → Kling 3.0 Turbo
Multi-shot storytelling (
multi_prompt, up to 6 shots) is unique to Kling 3.0 Turbo among these three models. If you are building a mini commercial, a travel reel, or a product story arc where scenes connect, Kling Turbo generates the full sequence in one pass at 30 cr/s (720p). Compare that to generating individual clips with Seedance or HappyHorse at 40 cr/s each and splicing in post — you save credits and time.
8.2You Need Storyboard Drafts or 480p Previews → Seedance 2.0 Mini
At 20 cr/s, Seedance 2.0 Mini at 480p delivers the lowest per-second credit rate of all three models. Generate 10 idea drafts at 5 seconds each for 1,000 credits — the same budget that buys roughly three 5-second HappyHorse 1.1 clips at 1080p (325 cr each). Once a concept is approved, re-generate the winner at 720p or switch to a higher-tier model for the final output.
8.3You Need Reference Video or Audio to Constrain the Style → Seedance 2.0 Mini
Only Seedance 2.0 Mini accepts video references and audio references as R2V inputs. If you have a reference clip whose motion rhythm or visual style you want the generation to mirror, Seedance Mini R2V is the only option among these three. HappyHorse 1.1 R2V is image-only; Kling Turbo has no R2V at all.
8.4You Need Synchronized Lip-Sync or Multilingual Dialogue → HappyHorse 1.1
HappyHorse 1.1 is Alibaba's model purpose-built for joint audio generation, including multilingual synchronized lip-sync. If your script requires a character speaking on screen in any language — English, Mandarin, Hindi, Spanish — HappyHorse 1.1 is the only option among these three that generates audio and lip motion in the same pass. Kling Turbo and Seedance Mini both generate audio, but without synchronized lip-sync.
8.5You Need 9-Image Character Consistency → HappyHorse 1.1
With up to 9 reference images addressed via
[Image 1] through [Image 9] in the prompt, HappyHorse 1.1 R2V offers the highest image-reference capacity of the three models. Use this for maintaining a specific character's appearance, wardrobe, and environment across multiple shots — a critical feature for branded video content, serialized social media, or narrative shorts.
8.6You Need Unusual Aspect Ratios (21:9, 9:21, 5:4, 4:5) → HappyHorse 1.1
HappyHorse 1.1 supports 9 aspect ratios including ultrawide 21:9 and ultra-tall 9:21. Kling 3.0 Turbo covers 3 (16:9, 9:16, 1:1) and Seedance 2.0 Mini covers 6 (adding 4:3, 3:4, 21:9 but not 9:21, 5:4, or 4:5). If your delivery target is a cinema screen (21:9) or a non-standard vertical format (5:4, 4:5), HappyHorse 1.1 is the only option.
9.Summary Table: Which Model Wins Each Dimension
| Dimension | Winner | Notes |
|---|---|---|
| Lowest cr/s @ 480p | Seedance 2.0 Mini | 20 cr/s — only model with 480p |
| Lowest cr/s @ 720p | Kling 3.0 Turbo | 30 cr/s vs 40 cr/s for the others |
| Lowest cr/s @ 1080p (Kie) | Kling 3.0 Turbo | 35 cr/s vs 65 cr/s for HappyHorse 1.1 |
| Lowest cr/s @ 1080p (Fal) | Kling 3.0 Turbo | 35 cr/s vs 45 cr/s for HappyHorse 1.1 |
| Multi-shot storytelling | Kling 3.0 Turbo | Up to 6 connected shots |
| Duration flexibility | Kling 3.0 Turbo | 3 s through 15 s in 1 s steps |
| Multimodal references (img+vid+aud) | Seedance 2.0 Mini | Only model supporting video + audio refs |
| Lip-sync audio | HappyHorse 1.1 | Native multilingual joint audio |
| Max reference images | HappyHorse 1.1 | Up to 9 images in R2V |
| Aspect ratio variety | HappyHorse 1.1 | 9 ratios including 21:9, 9:21, 5:4, 4:5 |
10.How to Access All Three Models on Imagera AI
All three models are available in the Imagera AI Video Generator. Select your model from the dropdown, choose your route (Normal = Kie route, Ultra Fast = Fal route), set resolution and duration, and generate. Credits are deducted per second of output at the rates shown in this guide.
For post-generation workflows:
- Smooth any output to 60fps with the Frame Interpolator
- Upscale resolution with the Video Enhancer
- Add a custom soundtrack with the Music Generator
Open the Imagera Video Generator →
11.Related Resources
- AI Frame Interpolator — Smooth any generated video to 60fps
- AI Video Enhancer — Upscale and denoise AI video output
- AI Frame Interpolation Guide — Full guide to RIFE-based frame interpolation
- AI Video Upscaler Comparison — Upscale video resolution after generation
- Imagera AI Video Generator — Generate video with all three models


