12 AI Video Models — One Subscription
Ten text-to-video engines — HappyHorse 1.0, Sora 2 Pro, Kling 3.0, Veo 3.1, Luma Ray 2, Hailuo 2.3, Seedance 2.0, Wan 2.6, Kling 2.5 Turbo, and Pika 2.2 — plus two avatar models (HeyGen V3 + Omnihuman, launching soon). All share a single credit pool on gVideo.
The roster
Every model, one credit pool
Alibaba's flagship — #1-ranked on Artificial Analysis blind tests for both text-to-video and image-to-video. Native audio synthesis, multimodal references, 50+ visual styles.
The most lifelike motion and strongest character consistency on the platform. Best for portrait, product and narrative shots.
Alibaba's fast, low-cost all-rounder. Handles everything from natural scenes to abstract visuals — ideal for high-volume iteration when stylization isn't the priority.
Photorealistic output with optional synced audio. The go-to when you need cinematic realism and diegetic sound.
Tight prompt adherence at a sensible credit cost. The best balance of quality and speed, and available on the free tier.
OpenAI's flagship Sora 2 Pro. Top-tier cinematic realism for hero shots and trailers. Pick Standard 720p (100 cr/4s) or HD 1080p (164 cr/4s). A cheaper non-Pro Sora 2 (720p) is also available.
The fast Turbo variant of Kling. Keeps Kling's signature motion quality while cutting render time and credit cost per second.
MiniMax Hailuo's third-gen with native audio. Standard 768p (30 cr/6s) or Pro 1080p (48 cr/6s) — the cheapest audio-included model on gVideo.
Luma's Ray 2 from the Dream Machine family. Fluid, dreamy motion for stylized shots. Standard (35 cr/5s) or HD 720p (60 cr/5s).
Pika's flagship text-to-video with signature stylized motion. Standard 720p (20 cr/5s) or HD 1080p (45 cr/5s) — the cheapest text-to-video on gVideo.
Turn a portrait photo plus a short script into a broadcast-quality talking avatar with built-in TTS. ~90 credits per 30-second render. Launching in the gVideo Avatar Studio.
Drive whole-body avatar motion from a photo and an audio track — head, torso, and gestures synced to speech. ~360 credits per 30-second render. Launching in the gVideo Avatar Studio.
Side by side
Model comparison at a glance
| Model | Short-run credits | Speed | Max length | Best for |
|---|---|---|---|---|
| Kling 3.0 | 40/ 5s | 60–120s | 10s | Lifelike motion · character consistency |
| Wan 2.6 | 30/ 5s | 30–75s | 5s | Fast · versatile |
| Veo 3.1 | 44/ 4s | 90–180s | 8s | Photorealism · optional audio |
| Veo 3.1 Lite | 20/ 4s | 60–120s | 8s | Cheapest audio-native · Google Veo |
| Veo 3.1 Pro | 140/ 4s | 120–240s | 8s | Top-tier cinematic hero shots · 4K Ultra |
| Seedance 2.0 | 90/ 5s | 40–90s | 10s | Speed · prompt adherence · free tier |
| Sora 2 Pro | 164/ 4s | 90–180s | 20s | OpenAI cinematic · HD default |
| HappyHorse 1.0 | 60/ 5s | 90–150s | 15s | #1 benchmark · native audio · 720p default |
| Kling 2.5 Turbo | 25/ 5s | 30–60s | 10s | Fast Kling renders |
| Hailuo 2.3 | 30/ 6s | 40–90s | 10s | Chinese realism · 768p default |
| Pika 2.2 | 20/ 5s | 30–75s | 10s | Stylized · 720p default |
| Luma Ray 2 | 45/ 5s | 40–90s | 9s | Dream Machine motion |
Speed ranges are indicative and fluctuate with provider load. Credits are identical on every plan tier.
Why aggregate
Twelve separate subs cost more and ship less
Running HappyHorse, Sora 2, Kling, Veo, Luma, Seedance, Hailuo, Wan, Pika, Kling Turbo, HeyGen, and Omnihuman directly on their provider sites means twelve dashboards, twelve credit meters, twelve billing cycles, and no way to compare takes against the same prompt.
gVideo consolidates the access: one prompt, one credit pool, every engine you want to try. When a shot isn’t landing on Wan, switch to Kling without starting a new session. When Veo’s queue is slow, fall back to Seedance. That flexibility is the entire point.
Credits are identical across every paid tier — you’re picking a budget, not a model. The model picker is always unlocked.
Common questions
Do I need separate subscriptions for each model?
No. gVideo is an aggregator — one subscription gives you access to HappyHorse 1.0, Sora 2 Pro, Kling 3.0, Veo 3.1, Luma Ray 2, Hailuo 2.3, Seedance 2.0, Wan 2.6, Kling 2.5 Turbo, and Pika 2.2, all drawing from a single monthly credit pool. Switch models anytime without any extra contract.
Which model should I start with?
For most first-time generations we recommend Wan 2.6 (cheapest) or Seedance 2.0 (best free-tier quality). Move to Kling 3.0 or Veo 3.1 once your prompt is locked and you need the highest fidelity.
Can I compare the same prompt across all twelve models?
Yes. Each model page on gVideo includes a side-by-side 'same prompt, cross-model' section. In the generator, you can also re-run the exact same prompt on another model in a single click.
Are these the official models from OpenAI, Kuaishou, Google, ByteDance, Alibaba, MiniMax and Luma?
Yes. gVideo integrates the official provider APIs. You are always running on the latest production version of each model, not a re-hosted checkpoint.
What is the difference in credit cost between models?
Defaults (cheapest tier × each model's shortest allowed clip): Pika 2.2 20 cr/5s, Kling 2.5 Turbo 25 cr/5s, Hailuo 2.3 Standard 30 cr/6s, Wan 2.6 30 cr/5s, Luma Ray 2 Standard 45 cr/5s, Kling 3.0 40 cr/5s, Veo 3.1 44 cr/4s, Seedance 2.0 90 cr/5s, Sora 2 (standard) 36 cr/4s, Sora 2 Pro HD 164 cr/4s. Sora, Hailuo, Pika, Luma, and Seedance offer additional resolution tiers at higher cost; full table on the pricing page.
Can I use the generated videos commercially?
Yes on all paid plans. The free tier is for personal use only. Paid plans include full commercial rights.