Text to video AI
No reference, no storyboard, no footage. Describe the shot in words and let the model decide the frame — the fastest way from an idea to something moving on screen.
At a glance
- Models for this job
- 43
- Free to run
- 14
- Paid from
- 60 cr a run
- Top of the range
- 525 cr a run
- Free account
- 22 credits a day, no card
Text-to-video gives the model everything to decide, which is both the appeal and the risk: you get shots you would never have framed yourself, and occasionally you get six fingers holding a coffee cup that pours upward.
The practical workflow is to draft on a cheap or free model until the wording is right, then spend the credits on a flagship engine once the prompt is doing what you meant.
The line-up
43 models do this job.
Read straight from the live catalogue — cost per run, what it takes as input, and what it can return.
| Model | Cost | Input | Length |
|---|---|---|---|
| Hailuo 2.3 | Free | Text prompt | 5s / 10s |
| Kling 2.1 Master | Free | Text prompt | 5s / 10s |
| Kling 3.0 | Free | Text prompt | 5s / 10s |
| Kling Pro | Free | Text prompt | 5s / 10s |
| Seedance 1.5 Pro | Free | Text prompt | 5s / 10s |
| Seedance 2.0 | Free | Text prompt | 5s / 10s |
| Seedance 2.0 Fast | Free | Text prompt | 5s / 10s |
| Seedance 2.0 Mini | Free | Text prompt | 5s / 10s |
| Seedance Lite | Free | Text prompt | 5s / 10s |
| Seedance Pro | Free | Text prompt | 5s / 10s |
| Veo 3.1 | Free | Text prompt | 4s / 6s / 8s |
| Veo 3.1 Fast | Free | Text prompt | 4s / 6s / 8s |
| Veo 3.1 Lite | Free | Text prompt | 4s / 6s / 8s |
| Wan 2.6 | Free | Text prompt | 5s / 10s |
| Flux 3 | 60 cr | Text prompt | 5s / 10s |
| Grok Imagine Video 1.5 | 60 cr | Text prompt | 5s / 10s |
| Happyhorse 1.1 | 60 cr | Text prompt | 5s / 10s |
| Luma Ray Flash 2 | 60 cr | Text prompt | 5s / 10s |
| Midjourney Video | 60 cr | Text prompt | 5s / 10s |
| Seedance 2 Fast | 60 cr | Text prompt | 4s / 5s / 6s / 8s / 10s / 12s |
| Seedance 2 Mini | 60 cr | Text prompt | 4s / 5s / 6s / 8s / 10s / 12s |
| Veo 3.1 Fast | 60 cr | Text prompt | 4s / 6s / 8s |
| Wan 2.6 Flash | 60 cr | Text prompt | 5s / 10s |
| Hailuo 03 | 140 cr | Text prompt | 5s / 10s |
| Hailuo 2.3 | 140 cr | Text prompt | 5s / 10s |
| Kling 2.5 Turbo Pro | 140 cr | Text prompt | 5s / 10s |
| Kling 2.6 | 140 cr | Text prompt | 5s / 10s |
| Kling 3.0 | 140 cr | Text prompt | 5s / 10s / 15s |
| Luma Ray 2 | 140 cr | Text prompt | 5s / 10s |
| Runway Gen4.5 | 140 cr | Text prompt | 5s / 10s |
| Seedance 2 | 140 cr | Text prompt | 4s / 5s / 6s / 8s / 10s / 12s |
| Seedance 2.5 | 140 cr | Text prompt | 4s / 5s / 8s / 10s / 12s / 15s / 20s / 25s / 30s |
| Wan 2.6 | 140 cr | Text prompt | 5s / 10s / 15s |
| Wan 2.7 | 140 cr | Text prompt | 5s / 10s / 15s |
| Wan 2.7 R2V | 140 cr | Text prompt | 5s / 10s |
| Wan 3.0 | 140 cr | Text prompt | 5s / 10s / 15s / 20s / 25s / 30s |
| Pika | 170 cr | Text prompt | 5s |
| Kling 3.0 Omni | 400 cr | Text prompt | 5s / 10s / 15s |
| Sora 2 | 400 cr | Text prompt | 4s / 8s / 10s / 12s / 15s |
| Sora 2 Pro | 400 cr | Text prompt | 4s / 8s / 10s / 12s / 15s |
| Veo 3.1 | 400 cr | Text prompt | 4s / 6s / 8s |
| Tencent Hunyuan | 425 cr | Text prompt | Model default |
| Kling O3 Pro | 525 cr | Text prompt | 5s |
How it works
Three steps.
Write the shot
Subject, action, camera, light. Short and specific beats long and poetic.
Draft cheap, finish premium
Fast clips exist so you can iterate on wording without burning a flagship budget.
Cut it together
Generate each shot, then assemble. These models make shots, not sequences.
Worth knowing
The honest small print.
- Expect to iterate on wording — the same prompt can read very differently to two models.
- Text on screen, hands and readable logos are still the weakest areas across every engine.
- Premium and flagship clips need a paid plan; the fast and free tiers are open to everyone.
- Credits are reserved when a run starts and settled when it ends, so a failed render is refunded.
FAQ
What people ask.
For dialogue and native audio, Veo. For camera control and consistency, Kling. For stylised speed, Seedance and Hailuo. For long establishing shots, whichever model on this page offers 10-second durations. Because they are all on one balance here, the sensible approach is to draft on a fast model and commit credits to a flagship only once the prompt is right.
Name the subject, the action, the camera move and the light, in that order, and keep it to a couple of sentences. Adding "cinematic, 8k, masterpiece" does very little; specifying "slow dolly in, late afternoon side light, shallow depth of field" does a lot. If a shot keeps failing, simplify the motion rather than adding adjectives.
Yes, on the free experimental models, which cost no credits. They are slower, they share capacity, and settings are requested rather than guaranteed — good for finding out whether a prompt works before paying for the version you keep.
Every model has a budget of attention. Long prompts with several competing instructions usually lose the last ones. Cut the brief to one subject and one motion, generate, then add a second element only once the first is holding.
Next door
Related tools.
Try it on your own brief
Free account, 22 credits a day, no card. Every model on this page is in the picker from the first render.